board #11389

# Conflicts:
#	README.md
#	docs/diagrams/cli-terminal.svg
#	docs/diagrams/promise-pillars.svg

AGENTS.md

@@ -46,7 +46,7 @@ Repository map and Reference Documentation sections below.
 
 ## Project at a Glance
 
-**OmniRoute** — unified AI proxy/router. One endpoint, 353 LLM providers, auto-fallback.
+**OmniRoute** — unified AI proxy/router. One endpoint, 351 LLM providers, auto-fallback.
 
 | Layer         | Location                | Purpose                                                                                                                                                                   |
 | ------------- | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

Dockerfile.bun

@@ -1,5 +1,5 @@
 # ── Multi-stage Dockerfile for Native Bun Runtime (web-latest-bun) ───────────
-FROM oven/bun:1.3.14-slim AS base
+FROM oven/bun:1.4.0-slim AS base
 WORKDIR /app
 
 RUN apt-get update \
@@ -50,7 +50,7 @@ ENV NODE_ENV=production
 RUN bun run --quiet build
 
 # ── Runner Base stage (100% Bun Native Production Runtime) ──────────────────
-FROM oven/bun:1.3.14-slim AS runner-base
+FROM oven/bun:1.4.0-slim AS runner-base
 
 LABEL org.opencontainers.image.title="omniroute" \
   org.opencontainers.image.description="Unified AI proxy — route any LLM through one endpoint (Bun Native)" \

README.md

@@ -7,7 +7,7 @@
 
 # 🚀 OmniRoute — The Free AI Gateway
 
-<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 353 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 353 AI providers · 90+ free tiers · ~1.51B free tokens/mo · 19 routing strategies · $0 to start."/>
+<img src="./docs/diagrams/readme-hero.svg" width="100%" alt="OmniRoute — Never stop coding. Every AI tool → 351 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot & Antigravity into FREE Claude / GPT / Gemini with auto-fallback. RTK + Caveman stacked compression saves 15–95% tokens (~89% avg) — never hit limits. 351 AI providers · 90+ free tiers · ~1.51B free tokens/mo · 19 routing strategies · $0 to start."/>
 
 </div>
 
@@ -210,7 +210,7 @@ curl http://localhost:20128/v1/chat/completions \
 
 </div>
 
-<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 353 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 353 providers · up to 95% token savings on eligible workloads · $0 to start with 90+ free tiers and 56 recurring/keyless free-forever providers · 35 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>
+<img src="./docs/diagrams/promise-pillars.svg" width="100%" alt="The Promise — One endpoint and 351 providers. Automatic fallback keeps routing while another healthy target is available. Six pillars: resilient fallback across 351 providers · up to 95% token savings on eligible workloads · $0 to start with 90+ free tiers and 56 recurring/keyless free-forever providers · 35 CLI/agent integrations through one config · OpenAI, Claude, Gemini and Responses API compatibility at /v1 · production controls including circuit breakers, TLS stealth, MCP 110 tools, A2A, memory, guardrails, evals and 39,000+ static test declarations across 5,100+ tracked test files."/>
 
 <br/>
 <br/>
@@ -461,7 +461,7 @@ All **19** strategies — mix & match per combo step:
 
 </div>
 
-<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 353 providers, 90+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 43 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>
+<img src="./docs/diagrams/comparison-table.svg" width="100%" alt="What sets OmniRoute apart — a dated feature snapshot vs 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute: 351 providers, 90+ free tiers built in, 19 routing strategies, 12-engine token compression, built-in MCP server with 110 tools, A2A agent protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, Desktop/Termux/PWA and 43 i18n UI locales. OmniRoute is MIT-licensed and self-hostable. Competitor capabilities and counts may change; see the linked methodology."/>
 
 <sub>📊 Full methodology &amp; per-feature detail vs 9router, OpenRouter, CLIProxyAPI &amp; LiteLLM → [`docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md`](docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md)</sub>
 
@@ -559,7 +559,7 @@ the current catalog at **[radar.omniroute.online/planos](https://radar.omniroute
 - **🖼️ New endpoints** — `/v1/ocr` (Mistral OCR) and `/v1/audio/translations` (Whisper-style) round out the media surface. → [API Reference](docs/reference/API_REFERENCE.md)
 - **🎨 Image / video / audio generation** — one API for media: xAI Grok Imagine & Novita AI video, ComfyUI, Freepik, Adobe Firefly, Microsoft Designer, Segmind, EdgeTTS. → [API Reference](docs/reference/API_REFERENCE.md)
 - **🌍 Deployment & ops** — reverse-proxy `basePath`, browser-language auto-detect, per-key device tracking, root-less MITM trust, zh-TW localization. → [Environment](docs/reference/ENVIRONMENT.md)
-- **🤝 More providers & agents** — Cursor Cloud Agent, Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **353-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
+- **🤝 More providers & agents** — Cursor Cloud Agent, Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **351-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
 - **📡 Routing transparency** — every response carries an `X-OmniRoute-Decision` header naming the strategy/provider/latency that served it, a new `cache-optimized` combo strategy + Auto-Combo `cacheAffinity` factor route repeat requests back to the connection holding the cached prefix, and a read-only `/v1/auto-combo/{channel}/candidates` endpoint exposes an `auto/*` channel's live candidate pool. → [Auto-Combo](docs/routing/AUTO-COMBO.md)
 - **⚡ Local performance & infra** — one-click local Redis, Cloudflare Workers / Deno Deploy relay deployers, Bifrost & Mux as supervised embedded services. → [Embedded Services](docs/frameworks/EMBEDDED-SERVICES.md)
 
@@ -642,11 +642,11 @@ of your shell history. → [CLI Integrations](docs/guides/CLI-INTEGRATIONS.md)
 
 <div align="center">
 
-## 🌐 353 AI Providers — 154 Catalog-Marked Free
+## 🌐 351 AI Providers — 154 Catalog-Marked Free
 
 </div>
 
-> **353 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **154 carrying `hasFree: true` discovery metadata**. The chat model registry covers **268 providers / 2,566 distinct provider-model pairs / 1,312 raw model IDs**; the separate free-budget catalog has **455 per-model rows**, **40 recurring pools** and **56 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).
+> **351 registered providers** across the canonical chat, media, search, local, cloud-agent and system collections, including **154 carrying `hasFree: true` discovery metadata**. The chat model registry covers **268 providers / 2,566 distinct provider-model pairs / 1,312 raw model IDs**; the separate free-budget catalog has **455 per-model rows**, **40 recurring pools** and **56 recurring/keyless free-forever providers**. These are different denominators by design; definitions and pool-deduped calculations live in the [Provider Reference](docs/reference/PROVIDER_REFERENCE.md) and [Free Tiers](docs/reference/FREE_TIERS.md).
 
 <div align="center">
 

changelog.d/features/11146-subscription-first-routing.md

@@ -1 +0,0 @@
-- **feat(routing):** subscription-first auto groupings — `auto/subscription` routes only through plan-included connections with a documented hard-stop overage and fails closed on exhaustion, while `auto/thrifty` orders the pool `subscription → keyless → free → cheap → premium` and steps up one rung at a time as each is exhausted. Billing class comes from a curated per-connection catalog (uncurated is treated as metered, never plan-included), both reuse STRICT_ZERO_COST's per-connection verification, and a quota reading whose `resetAt` has passed is now refreshed regardless of TTL so routing returns to plan capacity as soon as it resets ([#11146](https://github.com/diegosouzapw/OmniRoute/pull/11146))

changelog.d/fixes/11368-create-connection-tokenexpiresat.md

@@ -1 +0,0 @@
-- **Provider connections:** keep `tokenExpiresAt` when a connection is created. The create-path allowlist omitted it, so every insert stored NULL and the dashboard token badge could read a fresh connection as expired until its first background refresh ([#11368](https://github.com/diegosouzapw/OmniRoute/pull/11368)).

changelog.d/fixes/7592-windows-packaged-smoke.md

@@ -0,0 +1 @@
+- **fix(electron):** the packaged Windows desktop build now passes the #7592 cold-restart smoke end to end. Five defects were found and fixed on the way: (1) optional-pack staging failed on any machine whose `tar` is GNU tar — it reads the drive letter in an absolute `-f C:\...` archive path as a remote rsh host (`Cannot connect to C:`), so staging now passes a bare filename with `cwd` at the tarball directory and surfaces tar stderr on failure; (2) the Electron `files` allowlist was missing `lib/loginHeaderCapture.js`, so the packaged main process crashed at startup with `Cannot find module './lib/loginHeaderCapture'`; (3) electron-builder ≥26 injects an `!**/node_modules/**` ignore into every extraResources pattern list that no positive filter can override, which silently dropped the staged runtime `node_modules` (including the better-sqlite3 N-API prebuild) from `resources/app` — a fresh v3.8.50 desktop build would have shipped with no native SQLite driver and reproduced the #7592 sql.js fallback on every machine; an `afterPack` hook now restores the staged `node_modules` after packing; (4) the packaged-app smoke harness redirected `USERPROFILE` into its temp DATA_DIR, but Electron resolves the Roaming profile from `%USERPROFILE%\AppData\Roaming\<name>` (USERPROFILE takes precedence over the APPDATA env var) and the path service throws instead of creating it, so `requestSingleInstanceLock()` returned false and the app exited(0) silently before `app.whenReady()` — the harness now pre-creates the derived tree, and `ensureSmokeEnvDirs` is exported and unit-tested; (5) the #7592 driver assertion parsed a `[DB] Driver: ...` line that the server's primary DB path never printed (only the unused `openDatabaseAsync()` did) — `getDbInstance()` now emits the same line on open so the guard can actually assert the native driver. Also: the smoke env-allowlist unit test hardcoded POSIX paths and could only pass on Linux/macOS; expectations are now host-agnostic, with new regression tests for the USERPROFILE-derived tree and `tarPack` under absolute Windows-style paths

changelog.d/fixes/memory-embedding-quick-select-providers.md

@@ -0,0 +1 @@
+- **fix(memory):** Embedding Model Quick select, Embedding Source remote dropdown, and Rerank selector now list every configured provider with embedding/rerank support instead of only chat-catalog text matches plus OpenRouter live discovery; a generic OpenAI-compatible `/embeddings` + Cohere-compatible `/rerank` runtime fallback resolves any configured chat provider's embedding/rerank endpoint, so unlisted providers no longer fail with "Unknown embedding provider"; both memory selectors gained a free-text model override

config/quality/dependency-allowlist.json

@@ -12,6 +12,7 @@
     "@dnd-kit/core",
     "@dnd-kit/sortable",
     "@dnd-kit/utilities",
+    "@eslint/compat",
     "@huggingface/transformers",
     "@lobehub/icons",
     "@modelcontextprotocol/sdk",
@@ -60,6 +61,7 @@
     "eslint",
     "eslint-config-next",
     "eslint-plugin-sonarjs",
+    "espree",
     "express",
     "fast-check",
     "fetch-socks",

config/quality/eslint-suppressions.json

@@ -2372,16 +2372,6 @@
       "count": 2
     }
   },
-  "src/lib/guardrails/videoBridgeContactSheet.ts": {
-    "@typescript-eslint/no-unused-vars": {
-      "count": 1
-    }
-  },
-  "src/lib/guardrails/videoBridgeRuntime.ts": {
-    "@typescript-eslint/no-unused-vars": {
-      "count": 1
-    }
-  },
   "src/lib/guardrails/visionBridgeHelpers.ts": {
     "@typescript-eslint/no-unused-vars": {
       "count": 1
@@ -2422,11 +2412,6 @@
       "count": 1
     }
   },
-  "src/lib/memory/embedding/index.ts": {
-    "no-restricted-imports": {
-      "count": 1
-    }
-  },
   "src/lib/memory/genericBackend.ts": {
     "@typescript-eslint/no-unused-vars": {
       "count": 1

config/quality/file-size-baseline.json

@@ -478,8 +478,7 @@
     "_rebaseline_2026_08_23_tip_drift_post_batch0823": "Tip drift after the 2026-08-23 merge wave: chatBodyAdmission.ts 1009->1118 (+109, gate count incl. +1) and auth.ts 3337->3344 (+7), both grown by merges already on origin/release/v3.8.50 (verified identical on the pristine tip) — not by the codex-appserver-hardening PR that carries this bump. Owner pre-authorized baseline bumps 2026-08-22.",
     "_rebaseline_2026_08_24_11355_cooldown_recovery_guards": "PR #11355 own growth: test/route.ts 1215->1237, +22 (startup crash-recovery guard: clearStaleCrashCooldowns() now parses the persisted rate_limited_until deadline and skips clearing rows still genuinely in the future, instead of clearing every non-terminal cooldown unconditionally). Cohesive fix at the existing test-route dispatch chokepoint alongside the #11141 probe builder. Covered by tests/unit/startup-stale-cooldown-recovery.test.ts + tests/unit/repro-zai-cooldown-cleared-by-connection-test.test.ts.",
     "src/lib/guardrails/videoBridgeRuntime.ts": 1009,
-    "_rebaseline_2026_08_24_video_bridge_fu02_fu07_sampler": "PRs #11344 (FU-02 one-frame scene-aware determinism) + #11381 (FU-07 opt-in segment_aware structural sampling) own growth: videoBridgeRuntime.ts <1000->1009, +9 (sum of both boarded together in the same merge-batch). #11344 adds the deterministic one-frame midpoint fallback + policyEffective=uniform report at the existing scene_aware seam; #11381 adds the bounded local-only FFmpeg structural pre-analysis pass (scene/freeze/blur/exposure/SI-TI) and its budget-reallocation logic. Covered by tests/unit/guardrails/videoBridgeSampler.test.ts, tests/unit/guardrails/videoBridgeFu07StructuralSampling.test.ts, tests/integration/video-bridge-sampler-ffmpeg.test.ts. Owner pre-authorized rebaseline for legitimate PR growth (2026-08-19 directive).",
-    "open-sse/services/autoCombo/virtualFactory.ts": 1130
+    "_rebaseline_2026_08_24_video_bridge_fu02_fu07_sampler": "PRs #11344 (FU-02 one-frame scene-aware determinism) + #11381 (FU-07 opt-in segment_aware structural sampling) own growth: videoBridgeRuntime.ts <1000->1009, +9 (sum of both boarded together in the same merge-batch). #11344 adds the deterministic one-frame midpoint fallback + policyEffective=uniform report at the existing scene_aware seam; #11381 adds the bounded local-only FFmpeg structural pre-analysis pass (scene/freeze/blur/exposure/SI-TI) and its budget-reallocation logic. Covered by tests/unit/guardrails/videoBridgeSampler.test.ts, tests/unit/guardrails/videoBridgeFu07StructuralSampling.test.ts, tests/integration/video-bridge-sampler-ffmpeg.test.ts. Owner pre-authorized rebaseline for legitimate PR growth (2026-08-19 directive)."
   },
   "_rebaseline_base_2026_08_10_proxyfetch": "Base-red fix (green-prs sweep, issue #9985): open-sse/utils/proxyFetch.ts 1207 > cap 1000 — new proxied-TLS fetch helper introduced by the Fal reference-image work. Owner-authorized quick rebaseline to green; structural slim tracked for v3.9.0.",
   "_rebaseline_2026_07_27_v3849_train2": "Merge-train 2 (7 PRs) — owner-approved 2026-07-27. Single entry: chatCore.ts 4955->5006 (#8595, Responses multi-turn image compaction before the context hard-reject). Genuine irreducible growth at the existing compaction chokepoint in handleChatCore — the PR adds a last-resort retry against the concrete budget plus the estimateFinalInputTokens helper, both wired at the pre-existing call site rather than a new branch. Covered by tests/unit/8560-responses-image-compaction.test.ts (4 tests).",
@@ -647,6 +646,5 @@
   "_rebaseline_2026_08_18_imageregistry_merge_train": "merge-train 2026-08-18 (owner-authorized, /merge-prs batch of 84): open-sse/config/imageRegistry.ts crossed the 1000-line new-file cap for the first time purely from combining three independent, already-legitimate provider registrations boarded in the same local merge-train — #10542 (aihorde optional-key image catalog), #10494 (gemini-web image generation), #10594 (freepik/magnific provider rename + validation). 996 on release tip -> 1019 on the train tip. Each PR individually adds a small, additive IMAGE_PROVIDERS registry entry at the existing chokepoint; none crosses the cap alone. Not modularized as part of this train's gate fix (out of scope for a merge reconciliation, not a feature change). Covered by each PR's own focused tests (aihorde-image-catalog/generation, gemini-web image tests, freepik/magnific provider tests).",
   "_rebaseline_2026_08_20_v3850_merge_train_batch1": "Merge-train batch1 (2026-08-19/20, 30 PRs boarded onto release/v3.8.50): gateways.ts 1255->1268 = PR #10722 (Token Kiosk OpenAI-compatible provider gateway catalog entry, +13 declarative lines, same god-file no-split rationale as prior gateways.ts rebaselines); chatHelpers.ts (uncapped, not previously frozen) new 1017 = PR #10797 (relay/bifrost error normalization, +23/-2, own-PR growth, existing file already near cap from accumulated chokepoint wiring per its own rebaseline history above); chatBodyAdmission.ts (uncapped) new 1005 = pre-existing base-red on the pure release tip (1004>1000 before this train boarded anything, no PR in this batch touches this file) — frozen here at its current size, not authorizing further growth. Owner-authorized rebaseline (2026-08-19 merge-prs session).",
   "_rebaseline_2026_08_20_8338_cursor_image_provider": "PR (reimplementation of #8338, @valvesss): imageRegistry.ts 1019->1033 = new cursor IMAGE_PROVIDERS entry (Cursor plan image generation via Agent CLI), +14 lines of declarative provider metadata. Same god-registry no-split rationale as prior imageRegistry/gateways rebaselines.",
-  "_rebaseline_2026_08_20_imageregistry_1034": "imageRegistry.ts 1033->1034: +1 line drift between #10842 (cursor image provider, froze at 1033) and its actual merged state on release (measured 1034) — trivial rebaseline, not a new feature.",
-  "_rebaseline_2026_08_25_11146_subscription_first_auto": "PR #11146 (@yourspraveen, subscription-first auto groupings auto/subscription+auto/thrifty): open-sse/services/autoCombo/virtualFactory.ts is a NEW file in this PR landing at 1128 lines (+2 margin) — two opt-in flat auto ids built on the established auto/best-free pattern (connectionBillingCatalog + subscriptionLadder pure functions). Frozen at merge size per owner-authorized rebaseline directive (2026-08-19, merge-batch Step 4); no further growth without split rationale."
+  "_rebaseline_2026_08_20_imageregistry_1034": "imageRegistry.ts 1033->1034: +1 line drift between #10842 (cursor image provider, froze at 1033) and its actual merged state on release (measured 1034) — trivial rebaseline, not a new feature."
 }

docs/design/EXTERNAL-DB-BACKEND.md

@@ -0,0 +1,98 @@
+# Design: Pluggable External Durable-State DB Backend (PostgreSQL / MySQL)
+
+Status: PROPOSAL · Discussion issue: [#8075](https://github.com/diegosouzapw/OmniRoute/issues/8075) · Author: @oyi77
+
+Grounded in the tree at `release/v3.8.51` (`3192eb88d`). Every number below was measured on that tree, not estimated.
+
+---
+
+## 0. Why this document exists
+
+#8075 asks five questions (backends & order, abstraction contract, what stays SQLite-only, config surface, migration path). This document answers them with a census of how deeply the current data layer is coupled to SQLite, and proposes a repository-contract seam that lets external backends be adopted per domain instead of via a big-bang driver swap.
+
+## 1. Measured coupling census (`release/v3.8.51`)
+
+| Surface | Count | Implication for external backends |
+|---|---|---|
+| `src/lib/db/*.ts` modules calling `.prepare()` / `.transaction()` | **126 files** | The synchronous SQLite call shape is the de-facto data API; every module is a translation candidate |
+| `.pragma()` call sites | **36** | No equivalent in PG/MySQL — each needs an audit (many are performance tuning that becomes server config) |
+| SQLite migrations on disk | **159** | A second dialect needs its own migration lineage, not a translator |
+| `sqlite-vec` / `vec0` referencing files | **9** | Vector search must move to an external store (Qdrant proposed in #8075) |
+| FTS5 referencing files | **8** | Maps to Postgres `tsvector` / MySQL `FULLTEXT` — dialect work |
+| Existing driver adapters | **4** (`better-sqlite3`, `node:sqlite`, `bun:sqlite`, `sql.js`) behind `SqliteAdapter` (`src/lib/db/adapters/types.ts`) | Proves a seam exists at the DRIVER level — but all four are SQLite; the seam OmniRoute actually needs is one level up |
+
+The decisive fact: **the adapter interface is synchronous** (`prepare().run/get/all`, sync closure transactions). PostgreSQL and MySQL drivers are inherently async. A faithful external backend behind `SqliteAdapter` is therefore impossible without faking sync over async — which is exactly the trap #8037/#8073 already flagged ("not a drop-in ORM/driver swap"). The contract must live ABOVE the adapter layer.
+
+## 2. Proposed contract: durable-state repositories, not a driver swap
+
+Introduce a `DurableStateStore` port with per-domain repositories, defined once and implemented per dialect:
+
+```ts
+// src/lib/db/ports/ — dialect-agnostic, async, transactional by unit-of-work
+interface UnitOfWork {
+  connections: ConnectionRepository;   // provider connections + credentials
+  apiKeys: ApiKeyRepository;           // keys, groups, usage-limit fields
+  combos: ComboRepository;             // combo defs + routing policies
+  quotaState: QuotaStateRepository;    // quota/account state, cooldowns
+  sessionAffinity: AffinityRepository; // sticky routing state
+  audit: AuditRepository;              // config audit log
+}
+interface DurableStateStore {
+  readonly driver: "sqlite" | "postgres" | "mysql";
+  withUnitOfWork<T>(fn: (uow: UnitOfWork) => Promise<T>): Promise<T>;
+  healthCheck(): Promise<{ ok: boolean; latencyMs: number }>;
+}
+```
+
+Deliberate exclusions from v1 (stay SQLite-local, matching #8075 §3): WAL/VACUUM/PRAGMA lifecycle, file backup/import/export, `sqlite-vec` vector storage (external backends delegate vectors to Qdrant), FTS (dialect-native fulltext later).
+
+### What makes each repository shippable
+
+The single-writer pattern OmniRoute already uses internally (targeted raw-SQL stat bumps like `touchConnectionLastUsed`, single-writer caches) maps cleanly onto `UnitOfWork`. Repositories are adopted **per domain**: a deployment can run `postgres` for connections+keys while hot-path ephemeral state stays local — because the port is injected at the repository boundary, not under 126 modules.
+
+## 3. Sequencing: PostgreSQL first, MySQL second
+
+1. **Postgres-first** (`pg` + `pg-pool`): richer native JSONB (several modules persist JSON blobs), `tsvector`, `INSERT ... ON CONFLICT` semantics closest to `INSERT OR REPLACE` usage.
+2. **MySQL 8+ after**, gated by a shared **dialect-conformance suite**: the same behavioral test pack runs against both implementations (schema bootstrap, UoW rollback, credential round-trip incl. encryption-at-rest fields, pagination/ordering contracts). A backend is "supported" only when the suite is green.
+
+Rationale: one conformance suite prevents the MySQL implementation from drifting into a shim that passes smoke tests but corrupts ordering/collation edge cases.
+
+## 4. Config surface
+
+```
+DATABASE_DRIVER=sqlite            # sqlite (default) | postgres | mysql
+DATABASE_URL=postgres://...       # required when driver != sqlite
+DATABASE_POOL_MAX=10
+DATABASE_TLS_REJECT_UNAUTHORIZED=1
+DATABASE_MIGRATIONS_TABLE=_omniroute_migrations
+```
+
+- Default stays `sqlite` — zero-config npm/Electron/Termux behavior is non-negotiable (#8075 §6).
+- Health check surfaces in the existing `/api/monitoring/health` payload next to `inflightRequests`.
+- Migrations: per-dialect lineage directories (`migrations-postgres/`, `migrations-mysql/`) with their own runner sharing the safety checks (mass-migration abort, pre-migration backup → pg_dump/mysqldump equivalent) already proven in `migrationRunner.ts`.
+
+## 5. Migration path (one-way export tool)
+
+`omniroute db export --to postgres --dsn ...`:
+1. Snapshot `storage.sqlite` (reuse `createPreMigrationBackup`).
+2. Replay schema from the target dialect's lineage to head.
+3. Copy tables row-by-row through the new repositories (not raw SQL translation), preserving IDs; re-encrypt nothing (credential ciphertexts are dialect-neutral strings today).
+4. Verify with row-count + checksum comparison per table; write a receipt file.
+5. Refuse to continue on any mismatch (fail-closed).
+
+## 6. Phased delivery plan
+
+| Phase | Deliverable | Risk gate |
+|---|---|---|
+| P1 | `ports/` contract + SQLite implementation of `ConnectionRepository` + conformance-suite skeleton running against SQLite itself | Suite green on the reference driver |
+| P2 | Postgres implementation of the same repo + suite green on PG | Suite green on PG |
+| P3 | Remaining domains (apiKeys, combos, quotaState, affinity, audit) on PG | Full-suite parity vs SQLite run |
+| P4 | MySQL via conformance suite | Suite green on MySQL |
+| P5 | Export tool + docs | Round-trip receipt on a 1GB-scale fixture |
+
+P1–P2 alone deliver value: a clustered deployment can share connection/key state while everything else stays local.
+
+## 7. Explicitly out of scope
+
+- Translating the remaining ~120 SQLite-shaped modules verbatim ("just make prepare() hit Postgres") — rejected: it fakes sync over async and preserves dialect bugs forever.
+- Replacing `sqlite-vec` with pgvector — #8075 names Qdrant; keeping vector search out-of-process also keeps the lean proxy core lean (4.0 modular direction).

docs/diagrams/cli-terminal.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 350" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Animated terminal demoing the OmniRoute CLI: omniroute providers list (353 providers registered, anthropic, codex, glm, kimi shown active), omniroute combo list (always-on priority, cost-saver, fusion-panel, context-relay) and omniroute health (healthy, 18412 requests in 24h, p95 412ms, circuit breakers 24 closed, 1 half-open, 0 open), cycling over 85 top-level commands: providers, oauth, keys, combo, nodes, models, cache, compression, cost, usage, quota, health, resilience, telemetry, logs, audit, mcp, a2a, cloud, memory, skills, eval, doctor, repl, tunnel, backup, sync, webhooks, policy, pricing, translator, simulate and more.">
+<svg viewBox="0 0 1200 350" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Animated terminal demoing the OmniRoute CLI: omniroute providers list (351 providers registered, anthropic, codex, glm, kimi shown active), omniroute combo list (always-on priority, cost-saver, fusion-panel, context-relay) and omniroute health (healthy, 18412 requests in 24h, p95 412ms, circuit breakers 24 closed, 1 half-open, 0 open), cycling over 85 top-level commands: providers, oauth, keys, combo, nodes, models, cache, compression, cost, usage, quota, health, resilience, telemetry, logs, audit, mcp, a2a, cloud, memory, skills, eval, doctor, repl, tunnel, backup, sync, webhooks, policy, pricing, translator, simulate and more.">
 <desc>Compact animated terminal cycling three real OmniRoute CLI commands with a typewriter effect and a scrolling subcommand ticker; the first frame shows the completed providers-list screen.</desc>
 <defs><clipPath id="tickerClip"><rect x="12" y="304" width="1176" height="40"/></clipPath><clipPath id="tw0"><rect x="64" y="46" height="26" width="0"><animate attributeName="width" calcMode="discrete" values="0;31;61;92;122;153;184;214;245;245" keyTimes="0;0.012;0.018;0.024;0.030;0.036;0.042;0.048;0.054;1" dur="18s" repeatCount="indefinite"/></rect></clipPath><clipPath id="tw1"><rect x="64" y="46" height="26" width="0"><animate attributeName="width" calcMode="discrete" values="0;26;51;76;102;128;153;178;204;204" keyTimes="0;0.348;0.351;0.357;0.363;0.369;0.375;0.381;0.387;1" dur="18s" repeatCount="indefinite"/></rect></clipPath><clipPath id="tw2"><rect x="64" y="46" height="26" width="0"><animate attributeName="width" calcMode="discrete" values="0;20;41;61;82;102;122;143;163;163" keyTimes="0;0.678;0.684;0.690;0.696;0.702;0.708;0.714;0.720;1" dur="18s" repeatCount="indefinite"/></rect></clipPath></defs>
 <rect width="1200" height="350" fill="#0d1117"/>

docs/diagrams/comparison-table.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 780" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Comparison table: OmniRoute versus 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute is the only one with the full set: 353 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, a built-in MCP server with 110 tools, A2A protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, desktop/Termux/PWA, 43 UI locales and 100% MIT self-hosted. 9router has free providers, RTK compression and translation but no MCP, A2A, memory, guardrails, cloud agents or stealth. OpenRouter is a hosted SaaS with 400+ models, guardrails and a hosted MCP but is not self-hosted and lacks A2A, memory, cloud agents and stealth. CLIProxyAPI is a light OAuth proxy with two routing strategies. LiteLLM has 100+ providers, A2A and extensive guardrails but no memory, compression, free tier, stealth or cloud agents.">
+<svg viewBox="0 0 1200 780" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="Comparison table: OmniRoute versus 9router, OpenRouter, CLIProxyAPI and LiteLLM across 13 capabilities. OmniRoute is the only one with the full set: 351 providers, 90+ free providers built-in, 19 routing strategies, 12-engine token compression, a built-in MCP server with 110 tools, A2A protocol, persistent memory, guardrails, cloud agents, TLS fingerprint stealth, desktop/Termux/PWA, 43 UI locales and 100% MIT self-hosted. 9router has free providers, RTK compression and translation but no MCP, A2A, memory, guardrails, cloud agents or stealth. OpenRouter is a hosted SaaS with 400+ models, guardrails and a hosted MCP but is not self-hosted and lacks A2A, memory, cloud agents and stealth. CLIProxyAPI is a light OAuth proxy with two routing strategies. LiteLLM has 100+ providers, A2A and extensive guardrails but no memory, compression, free tier, stealth or cloud agents.">
   <desc>Static-header comparison table where each capability row fades in top to bottom; the OmniRoute column is highlighted and shows a check or a leading value in every row, while competitors show a mix of checks, partials and crosses.</desc>
   <defs>
     <pattern id="gC" width="32" height="32" patternUnits="userSpaceOnUse"><path d="M 32 0 L 0 0 0 32" fill="none" stroke="#ffffff" stroke-opacity="0.05" stroke-width="1"/></pattern>

docs/diagrams/promise-pillars.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 540" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="The OmniRoute promise: one endpoint and 353 providers. Six pillars. Resilient fallback: automatic routing continues while another healthy target is available. Save up to 95 percent of eligible tokens: RTK plus Caveman stacked compression averages about 89 percent on tool-heavy sessions. Zero dollars to start: 90+ providers with a free tier and 56 recurring or keyless free-forever providers. Every tool works: 35 CLI and agent integration records, including Claude Code, Codex, Cursor, Cline, Copilot and Antigravity, through one config. One endpoint: OpenAI, Claude, Gemini and Responses API translation at /v1. Production controls: circuit breakers, TLS stealth, MCP with 110 tools, A2A, memory, guardrails, evals, and 39,000+ static test declarations across 5,100+ tracked test files.">
+<svg viewBox="0 0 1200 540" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="The OmniRoute promise: one endpoint and 351 providers. Six pillars. Resilient fallback: automatic routing continues while another healthy target is available. Save up to 95 percent of eligible tokens: RTK plus Caveman stacked compression averages about 89 percent on tool-heavy sessions. Zero dollars to start: 90+ providers with a free tier and 56 recurring or keyless free-forever providers. Every tool works: 35 CLI and agent integration records, including Claude Code, Codex, Cursor, Cline, Copilot and Antigravity, through one config. One endpoint: OpenAI, Claude, Gemini and Responses API translation at /v1. Production controls: circuit breakers, TLS stealth, MCP with 110 tools, A2A, memory, guardrails, evals, and 39,000+ static test declarations across 5,100+ tracked test files.">
   <desc>Animated promise card: six pillar tiles fade in in reading order, then a soft colored border highlight sweeps from tile to tile in a continuous cycle.</desc>
   <defs>
     <pattern id="gridPaperP" width="32" height="32" patternUnits="userSpaceOnUse">
@@ -21,7 +21,7 @@
     <line x1="150" y1="53" x2="1160" y2="53" stroke="#232b38" stroke-width="1.5"/>
   </g>
   <g>
-    <text x="40" y="100" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="23" font-weight="600" fill="#c9d1d9">One endpoint. <tspan fill="#a78bfa" font-weight="800">353 providers.</tspan> Never stop building — OmniRoute picks <tspan fill="#7ee787" font-weight="700">the cheapest one that works</tspan>.</text>
+    <text x="40" y="100" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="23" font-weight="600" fill="#c9d1d9">One endpoint. <tspan fill="#a78bfa" font-weight="800">351 providers.</tspan> Never stop building — OmniRoute picks <tspan fill="#7ee787" font-weight="700">the cheapest one that works</tspan>.</text>
   </g>
 
   <g font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif">
@@ -38,7 +38,7 @@
         <line x1="3.9" y1="3.9" x2="18.1" y2="18.1"/>
       </g>
       <text x="102" y="170" font-size="18" font-weight="800" fill="#74b9ff">Never hit limits</text>
-      <text x="66" y="204" font-size="13.5" fill="#a1a1aa">Auto-fallback across 353 providers in</text>
+      <text x="66" y="204" font-size="13.5" fill="#a1a1aa">Auto-fallback across 351 providers in</text>
       <text x="66" y="226" font-size="13.5" fill="#a1a1aa">milliseconds. Quota out? The next provider</text>
       <text x="66" y="248" font-size="13.5" fill="#a1a1aa">takes over while a healthy target remains.</text>
     </g>

docs/diagrams/readme-hero.svg

@@ -1,4 +1,4 @@
-<svg viewBox="0 0 1200 548" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="OmniRoute hero: Never stop coding. Every AI tool to 353 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot and Antigravity into free Claude, GPT and Gemini with auto-fallback. RTK + Caveman stacked compression saves 15 to 95 percent of tokens — about 89 percent average on tool-heavy sessions — so you never hit limits. Stats: 353 AI providers, 90+ free tiers, about 1.51B free tokens per month, 15 to 95 percent token savings, 19 routing strategies, zero dollars to start.">
+<svg viewBox="0 0 1200 548" xmlns="http://www.w3.org/2000/svg" role="img" aria-label="OmniRoute hero: Never stop coding. Every AI tool to 351 providers — 90+ free — through one endpoint. Claude Code, Codex, Cursor, Cline, Copilot and Antigravity into free Claude, GPT and Gemini with auto-fallback. RTK + Caveman stacked compression saves 15 to 95 percent of tokens — about 89 percent average on tool-heavy sessions — so you never hit limits. Stats: 351 AI providers, 90+ free tiers, about 1.51B free tokens per month, 15 to 95 percent token savings, 19 routing strategies, zero dollars to start.">
   <desc>Animated hero card: a pulse travels the divider line and a compression bar demo repeatedly shrinks a prompt by up to 95 percent; all headline content is static and readable on the first frame.</desc>
   <defs>
     <pattern id="gridPaperH" width="32" height="32" patternUnits="userSpaceOnUse">
@@ -28,7 +28,7 @@
   <text x="48" y="138" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="60" font-weight="800" fill="#e9edf3">Never stop coding<tspan fill="#a855f7">.</tspan></text>
 
   <!-- subheadline -->
-  <text x="48" y="184" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="25" font-weight="600" fill="#c9d1d9">Every AI tool → <tspan fill="#a78bfa" font-weight="800">353 providers</tspan> — <tspan fill="#7ee787" font-weight="800">90+ free</tspan> — through one endpoint.</text>
+  <text x="48" y="184" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="25" font-weight="600" fill="#c9d1d9">Every AI tool → <tspan fill="#a78bfa" font-weight="800">351 providers</tspan> — <tspan fill="#7ee787" font-weight="800">90+ free</tspan> — through one endpoint.</text>
 
   <!-- plug line -->
   <text x="48" y="222" font-family="Inter, 'Segoe UI', Arial, Helvetica, system-ui, sans-serif" font-size="16.5" fill="#a1a1aa">Claude Code · Codex · Cursor · Cline · Copilot · Antigravity&#160;&#160;→&#160;&#160;<tspan fill="#7ee787" font-weight="700">FREE</tspan> Claude / GPT / Gemini · auto-fallback</text>

docs/guides/TIERS.md

@@ -24,11 +24,8 @@ it expires.
 | Antigravity / Devin Desktop         | Built-in quotas                              |
 
 **Strategy**: route here first for every request that fits the model's
-strengths. The quota tracker monitors approaching resets, and the `reset-aware`
-combo strategy prioritizes accordingly. To route Tier 1 first and only step out
-to paid tiers as quota runs out, use the `auto/thrifty` id — or `auto/subscription`
-to stay on plan-included capacity and fail closed instead. See
-[Subscription-first routing](../routing/SUBSCRIPTION_LADDER.md).
+strengths. Quota tracker monitors approaching reset; combo strategies
+`reset-aware` and `subscription` prioritize accordingly.
 
 ## Tier 2 — Cheap
 

docs/i18n/ar/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/az/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/bg/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/bn/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/cs/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/da/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/de/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/es/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/fa/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/fi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/fr/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/gu/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/he/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/hi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/hu/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/id/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/in/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/it/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ja/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ko/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/mr/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ms/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/nl/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/no/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/phi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/pl/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/pt-BR/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/pt/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ro/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ru/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/sk/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/sv/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/sw/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ta/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/te/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/th/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/tr/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/uk-UA/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/ur/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/vi/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/zh-CN/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/i18n/zh-TW/llm.txt

@@ -4,7 +4,7 @@
 
 ---
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -281,7 +281,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -479,7 +479,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

docs/reference/PROVIDER_REFERENCE.md

@@ -1,16 +1,16 @@
 ---
 title: "Provider Reference"
-version: 3.8.51
-lastUpdated: 2026-08-25
+version: 3.8.50
+lastUpdated: 2026-08-24
 ---
 
 # Provider Reference
 
 > **Auto-generated** from `src/shared/constants/providers.ts` — do not edit by hand.
 > Regenerate with: `npm run gen:provider-reference`
-> **Last generated:** 2026-08-25
+> **Last generated:** 2026-08-24
 
-Total providers: **353**. See category breakdown below.
+Total providers: **351**. See category breakdown below.
 
 ## Categories
 
@@ -122,7 +122,7 @@ Use the dashboard at `/dashboard/providers` to enable, configure, and test each
 | `zai-web` | `zw` | Z.ai Web | Web cookie | [link](https://chat.z.ai) | Copy the "token" value from chat.z.ai → DevTools → Application → Local Storage. Do not copy cookies; OmniRoute handles the per-request CAPTCHA through its browser transport. | — |
 | `zenmux-free` | `zmf` | ZenMux Free (Web) | Web cookie | [link](https://zenmux.ai) | Login at zenmux.ai, then export all cookies using EditThisCookie or Cookie-Editor and paste the full Cookie header string here. Refresh every ~30 days. | — |
 
-## API Key Providers (paid / paid-with-free-credits) (233)
+## API Key Providers (paid / paid-with-free-credits) (231)
 
 | ID | Alias | Name | Tags | Website | Notes |
 |----|-------|------|------|---------|-------|
@@ -342,8 +342,6 @@ Use the dashboard at `/dashboard/providers` to enable, configure, and test each
 | `vertex-partner` | `vp` | Vertex AI Partners | API key, enterprise | [link](https://cloud.google.com/vertex-ai) | Provide the same Service Account JSON used for Vertex AI partner models. |
 | `void-ai` | `void-ai` | Void AI | API key, aggregator | [link](https://voidai.app) | The public model catalog marks some models with a free plan requirement, but access is conditional and no numeric quota is confirmed. |
 | `volcengine` | `volcengine` | Volcengine | API key | [link](https://www.volcengine.com) | — |
-| `volcengine-agent-plan` | `veap` | Volcengine Ark Agent Plan | API key | [link](https://console.volcengine.com/ark/region:cn-beijing/subscription/agent-plan) | Connect your Volcano Engine account or use an Ark Agent Plan subscription API key. |
-| `volcengine-coding-plan` | `vecp` | Volcengine Ark Coding Plan | API key | [link](https://console.volcengine.com/ark/region:cn-beijing/subscription/coding-plan) | Connect your Volcano Engine account or use an Ark Coding Plan subscription API key. |
 | `voyage-ai` | `voyage` | Voyage AI | API key, embed/rerank | [link](https://www.voyageai.com) | Bearer API key for Voyage AI embeddings and rerank APIs. |
 | `wafer` | `wafer` | Wafer AI | API key | [link](https://wafer.ai) | — |
 | `wandb` | `wandb` | Weights & Biases Inference | API key | [link](https://wandb.ai) | — |

docs/routing/SUBSCRIPTION_LADDER.md

@@ -1,182 +0,0 @@
----
-title: "Subscription-first routing"
-version: 3.8.50
-lastUpdated: 2026-08-22
----
-
-# Subscription-first routing
-
-> Two new `auto/*` ids — `auto/subscription` and `auto/thrifty`. Both are opt-in by being
-> requested: nothing routes through them unless a caller asks for the id by name, and no
-> existing pool, strategy, or default changes.
-
-## Why this exists
-
-OmniRoute already answers two cost questions, and neither is the one most operators ask.
-
-| Existing mechanism                                       | Answers                             |
-| -------------------------------------------------------- | ----------------------------------- |
-| `hidePaidModels` (`autoCombo/paidModelFilter.ts`)        | "is this model catalogued free?"    |
-| `freeAccessPolicy: "strict"` (`strictZeroCostFilter.ts`) | "can this connection ever bill me?" |
-| `quotaPreflight` (`combo/quotaExhaustionCutoff.ts`)      | "is this connection near its wall?" |
-| `budgetCap` / `budgetFallback` (`autoCombo/engine.ts`)   | "cap spend, degrade to cheapest"    |
-
-Every free-only mechanism **fails closed** — an exhausted free pool is an empty pool, never a
-step up to a paid option — and every paid-side mechanism is tier-agnostic. Neither answers:
-
-> "Use the quota I already pay for. When it runs out, either stop, or step up one rung at a
-> time through the cheapest paid options — and come back the moment it resets."
-
-## Billing is a connection fact, not a model fact
-
-`classifyTier()` (`open-sse/services/tierResolver.ts`) keys on `(provider, model)` and returns
-`free | cheap | premium` from catalog pricing. But whether a request costs incremental money
-depends on **which connection serves it**: the same model is plan-included through a Claude Code
-OAuth connection and billed per token through an API-key connection.
-
-`provider_connections.auth_type` is not a safe proxy in either direction — metered OAuth
-connections exist, and plan-included API-key connections exist (a Copilot seat token is not a
-metered API key). So billing class comes from a **curated catalog**,
-`open-sse/config/connectionBillingCatalog.ts`, hand-set from each provider's published terms —
-the same pattern `FreeModelBudget.hardStopGuaranteed` already established for free models.
-
-```ts
-type ConnectionBillingClass = "subscription" | "metered" | "keyless" | "unknown";
-type ConnectionOverageBehavior = "hard-stop" | "meters-to-paid" | "unknown";
-```
-
-Resolution order (`autoCombo/connectionBilling.ts`): the synthetic no-auth sentinel →
-`keyless`; a catalog entry matching provider **and** `authType`; a provider-wide entry;
-otherwise `unknown`. **Uncurated is not free** — `unknown` is consumed as `metered`
-everywhere, so a provider added tomorrow starts outside the subscription rung and has to be
-curated in deliberately.
-
-## The rung model
-
-Five rungs in escalation order. They differ in more than price — each has its **own**
-exhaustion signal, which is why this is not merely a sort.
-
-| #   | Rung           | Membership                                         | Exhausted when                |
-| --- | -------------- | -------------------------------------------------- | ----------------------------- |
-| 0   | `subscription` | curated `billing: "subscription"`                  | quota window at/below cutoff  |
-| 1   | `keyless`      | the synthetic no-auth path                         | connection cooldown / breaker |
-| 2   | `free`         | metered connection, `classifyTier() === "free"`    | free allowance exhausted      |
-| 3   | `cheap`        | metered connection, `classifyTier() === "cheap"`   | per-rung budget consumed      |
-| 4   | `premium`      | metered connection, `classifyTier() === "premium"` | per-rung budget consumed      |
-
-Rungs 0-2 exhaust on **quota**, which is observable and already tracked. Rungs 3-4 have no
-quota — a paid connection serves forever — so their only sane exhaustion signal is a per-rung
-**budget**. Without one, "escalate when cheap is exhausted" has no trigger.
-
-## `auto/subscription` — fail closed
-
-Pool = rung 0 only, restricted to connections whose overage is a documented `hard-stop`, each
-verified live to have quota headroom. Everything ambiguous is excluded: an uncurated provider,
-an unverifiable quota reading, a stale reading, or an overage that meters to paid.
-
-An empty pool is the **intended** answer, not a defect — the caller's existing empty-pool path
-turns it into a clear error rather than a silent, billable fallback. That is the whole promise
-of the id.
-
-`keyless` deliberately does **not** qualify: this grouping means "the plan I pay for", so a
-no-auth backend does not belong in it. Use `auto/thrifty` (or `auto/best-free`) for that.
-
-### Connection safety
-
-A candidate is not always tied to one connection — a logical candidate carries an
-`allowedConnectionIds` allowlist, and the account actually used is chosen later, at dispatch,
-by `open-sse/services/combo/autoStrategy.ts`. Both groupings therefore verify **each connection
-individually** and rewrite `allowedConnectionIds` down to exactly the surviving subset — never
-the full original list, never one arbitrarily-chosen member. Because `autoStrategy.ts` already
-enforces that array as a hard allowlist, rewriting it here makes "verified" and "actually used"
-the same set by construction. This is the same invariant, and the same reasoning, as
-[STRICT_ZERO_COST](./STRICT_ZERO_COST.md).
-
-## `auto/thrifty` — escalate one rung at a time
-
-Pool = all rungs, ordered by rung index, with exhausted candidates gated out. The `auto` engine
-still scores **within** the surviving pool: the ladder decides which rungs are in play, scoring
-decides which candidate wins inside them. Ordering is stable within a rung, so the scorer's own
-ranking is never reshuffled by this overlay.
-
-This is an ordering + gating overlay, **not** a new dispatcher: `combo.ts`'s speculative loop
-already walks targets in order and falls through on failure, so a runtime exhaustion the
-preflight did not catch still escalates to the next rung inside the same request.
-
-Where `auto/subscription` fails closed, `auto/thrifty` fails **open**: a plan-included
-connection with no usable quota reading is still tried first. Trying it costs nothing, and if
-it turns out to be exhausted the fall-through reaches the next rung anyway — whereas refusing
-to try it would send the request to a paid rung on missing telemetry, the exact outcome the
-grouping exists to avoid.
-
-## Returning to the plan after a reset
-
-Three independent things must expire before routing returns to rung 0. Fixing only one leaves
-the ladder stuck on paid rungs long after the plan refilled.
-
-1. **The quota-state cache** — `freeAccessQuota.ts` caches per `(provider, connection)` with a
-   180s TTL. A cached entry whose own `resetAt` has already passed describes a window that no
-   longer exists, so it is now treated as stale **regardless of age** and forces a refresh.
-   Without this, a plan that refilled at midnight keeps reading exhausted until the TTL happens
-   to lapse.
-2. **The ladder's own state** — there is none, by design. Rung eligibility is recomputed from
-   live quota state on every pool build; no persisted "currently on rung 3" record exists that
-   could outlive a reset and wedge routing.
-3. **The connection cooldown** — the exhausting 429 sets `rateLimitedUntil` from exponential
-   backoff, which for a plan connection can overshoot the real reset. `clampCooldownToReset()`
-   (`subscriptionLadder.ts`) narrows a cooldown to the upstream's own reset instant and can
-   never extend one. **It is implemented and tested but not yet wired**: the quota cache is
-   invalidated in `src/sse/services/auth.ts` _before_ any cooldown is written, so `resetAt`
-   must be captured earlier in that function — a change to the resilience hot path that
-   belongs in its own reviewed PR. Until then, re-entry waits out the connection cooldown
-   (which already prefers upstream `Retry-After` hints when the provider sends them).
-
-### Anti-flap
-
-A rung that just reset is re-admitted only above `reentryMinRemainingPercent` (default 5),
-while a connection already in play only has to stay above `exitCutoffPercent` (default 2,
-matching `quotaPreflight.defaultThresholdPercent`). The gap is the hysteresis band — without
-it, a connection hovering at the cutoff oscillates between rungs on consecutive requests.
-
-## Configuration
-
-Tuning only. There is deliberately **no** `enabled` flag: a toggle able to switch these off
-would leave `auto/subscription` quietly serving the full pool — paid models included — under a
-name that promises the opposite.
-
-```jsonc
-{
-  "subscriptionLadder": {
-    "exitCutoffPercent": 2,
-    "reentryMinRemainingPercent": 5,
-    "rungBudgetUsd": { "cheap": 5.0, "premium": 0 }, // 0 disables a rung outright
-  },
-}
-```
-
-Budget gating is inert until a spend resolver is wired: with no accounting available a paid
-rung is ordered but never gated. As of v3.8.51 the `rungBudgetUsd` setting is accepted by the
-schema but NOT yet enforced — treat it as reserved configuration, not an active spend cap. Rung ordering, quota-based exhaustion, and reset re-entry all
-work without it.
-
-## Composition
-
-`subscription` and `thrifty` are `AutoTier` values, so they compose with every category:
-`auto/coding:thrifty`, `auto/reasoning:subscription`, and so on. The two flat ids
-(`auto/subscription`, `auto/thrifty`) are advertised in `/v1/models` and the dashboard.
-
-Neither id is paid-tier, so `isPaidTierAutoId()` returns `false` for both and
-`auto/subscription` survives `hidePaidModels`.
-
-## Where the code lives
-
-| Concern                         | File                                                |
-| ------------------------------- | --------------------------------------------------- |
-| Curated billing facts           | `open-sse/config/connectionBillingCatalog.ts`       |
-| Classifier                      | `open-sse/services/autoCombo/connectionBilling.ts`  |
-| Rungs, both groupings, re-entry | `open-sse/services/autoCombo/subscriptionLadder.ts` |
-| Wiring into the candidate pool  | `open-sse/services/autoCombo/virtualFactory.ts`     |
-| Reset-aware cache staleness     | `open-sse/services/autoCombo/freeAccessQuota.ts`    |
-| Tier surface                    | `open-sse/services/autoCombo/suffixComposition.ts`  |
-| Advertised ids                  | `open-sse/services/autoCombo/builtinCatalog.ts`     |
-| Tests                           | `tests/unit/autoCombo/subscription-ladder.test.ts`  |

electron/package.json

@@ -45,6 +45,7 @@
     "copyright": "Copyright © 2025 OmniRoute",
     "buildDependenciesFromSource": false,
     "npmRebuild": false,
+    "afterPack": "../scripts/build/afterpack-copy-node-modules.mjs",
     "directories": {
       "output": "dist-electron",
       "buildResources": "assets"
@@ -58,6 +59,7 @@
       "main.js",
       "preload.js",
       "loginManager.js",
+      "lib/loginHeaderCapture.js",
       "processTree.js",
       "sqlite-inspection.js",
       "remoteServerPromptPreload.js",

eslint.config.mjs

@@ -1,4 +1,6 @@
+import { fixupConfigRules } from "@eslint/compat";
 import nextVitals from "eslint-config-next/core-web-vitals";
+import * as espree from "espree";
 import tseslint from "typescript-eslint";
 
 // #7879: bar NEW local `toNumber` definitions outside the canonical helper.
@@ -37,7 +39,22 @@ const IMPORT_BOUNDARY_RESTRICTIONS = {
 
 /** @type {import("eslint").Linter.Config[]} */
 const eslintConfig = [
-  ...nextVitals,
+  ...fixupConfigRules(nextVitals),
+  // eslint-config-next's Babel parser has not adopted ESLint 10's ScopeManager
+  // finalize contract yet. Plain JS/JSX does not need that parser, so use the
+  // ESLint-native parser while retaining Next's plugins and rules.
+  {
+    files: ["**/*.{js,jsx,mjs,cjs}"],
+    languageOptions: {
+      parser: espree,
+    },
+  },
+  {
+    files: ["**/*.{mts,cts}"],
+    languageOptions: {
+      parser: tseslint.parser,
+    },
+  },
   // Pacote 4 (plano mestre testes+CI, 2026-07-04) — zero-warning policy: TODA regra roda
   // como "error" e a dívida pré-existente vive congelada por arquivo+regra em
   // config/quality/eslint-suppressions.json (ESLint bulk suppressions nativo). Violação
@@ -133,28 +150,45 @@ const eslintConfig = [
       "no-restricted-syntax": "off",
     },
   },
-  // Relaxed rules for open-sse and tests (incremental adoption)
+  // Relaxed rules for TypeScript in open-sse and tests (incremental adoption).
+  // eslint-config-next already registers @typescript-eslint for every TS file;
+  // registering a second plugin object is rejected by ESLint 10.
   {
-    files: ["open-sse/**/*.ts", "tests/**/*.mjs", "tests/**/*.ts"],
-    plugins: {
-      "@typescript-eslint": tseslint.plugin,
-    },
+    files: ["open-sse/**/*.ts", "tests/**/*.ts"],
     rules: {
       "@typescript-eslint/no-explicit-any": "error",
       "@next/next/no-assign-module-variable": "off",
       "react-hooks/rules-of-hooks": "off",
     },
   },
+  {
+    files: ["tests/**/*.mjs"],
+    plugins: {
+      "@typescript-eslint": tseslint.plugin,
+    },
+    rules: {
+      "@next/next/no-assign-module-variable": "off",
+      "react-hooks/rules-of-hooks": "off",
+    },
+  },
+  // JS/JSX files do not match eslint-config-next's TypeScript block. Register
+  // the plugin only for that disjoint scope so the shared unused-vars ratchet
+  // works without redefining the plugin for TS/TSX under ESLint 10.
+  {
+    files: ["src/**/*.{js,jsx}"],
+    plugins: {
+      "@typescript-eslint": tseslint.plugin,
+    },
+  },
   // Ratchet: bar NEW unused vars/args/catches outside the `_` escape hatch.
   // Pre-existing violations are frozen via config/quality/eslint-suppressions.json
   // (same pattern as #7879 toNumber); only genuinely NEW unused bindings fail
   // lint. `args: "all"` (not `after-used`) so a leading unused param is never
   // silently skipped, e.g. `function handle(req, _opts, next)` must flag `req`.
+  // eslint-config-next already registers @typescript-eslint; registering it again
+  // in this block is rejected by ESLint 10.
   {
     files: ["src/**/*.{ts,tsx,js,jsx}", "open-sse/**/*.ts", "tests/**/*.{ts,tsx,mjs}"],
-    plugins: {
-      "@typescript-eslint": tseslint.plugin,
-    },
     rules: {
       "@typescript-eslint/no-unused-vars": [
         "error",

llm.txt

@@ -1,6 +1,6 @@
 # OmniRoute
 
-> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 353 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
+> OmniRoute is a free, open-source AI Gateway that acts as a universal API proxy for multi-provider LLMs. It provides smart routing, automatic fallback, load balancing, and format translation across 351 AI providers — all through a single OpenAI-compatible endpoint. Includes a built-in MCP Server (110 tools), A2A v0.3 protocol, Memory/Skills systems, Cloud Agents (codex, cursor, devin, jules), Guardrails framework, and an Electron desktop app.
 
 ## Overview
 
@@ -277,7 +277,7 @@ OmniRoute solves the problem of managing multiple AI provider subscriptions, quo
 ## Key Features (v3.8.50)
 
 ### Core Proxy
-- **353 AI providers** with automatic format translation
+- **351 AI providers** with automatic format translation
 - **Provider categories**: Free (90+ free tiers), OAuth, API Key, Self-Hosted, Custom (OpenAI/Anthropic-compatible)
 - **18 routing strategies**: priority, weighted, round-robin, fill-first, p2c, random, least-used, cost-optimized, reset-aware, reset-window, headroom, strict-random, auto, lkgp, context-optimized, context-relay, fusion, pipeline
 - **4-tier fallback**: Subscription → API Key → Cheap → Free
@@ -475,7 +475,7 @@ diagnostics) plus **memory**, **skill**, **agentSkill**, **githubSkill**, **pool
 
 ## v3.8.x Highlights
 
-- **353-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
+- **351-provider catalog** with 90+ free tiers, one-click account imports, and bulk key add
 - **19 routing strategies** — including `fusion` (parallel panel + judge synthesis), `pipeline`, `reset-aware`, `reset-window`, `headroom`, and `context-relay`
 - **14-factor Auto-Combo scoring** with bandit exploration and progressive cooldown
 - **MCP server expanded to 110 tools / 33 scopes** (canonical + memory/skill/agentSkill/githubSkill/pool/notion/obsidian/localCorpus/gamification/plugin modules)

open-sse/config/connectionBillingCatalog.ts

@@ -1,143 +0,0 @@
-/**
- * Curated billing classification for provider CONNECTIONS.
- *
- * The economic tier resolver (`open-sse/services/tierResolver.ts`) answers
- * "how much does this MODEL cost per token?" — a fact about the catalog. It
- * cannot answer the question subscription-first routing actually needs:
- *
- *   "does serving this request through THIS connection cost incremental money,
- *    or is it already covered by a flat-rate plan the operator pays anyway?"
- *
- * That is a property of the credential, not the model: `claude/claude-*` is
- * plan-included through a Claude Code OAuth connection and billed per token
- * through an API-key connection — same provider, same model, opposite
- * economics.
- *
- * `provider_connections.auth_type` alone is NOT a safe proxy in either
- * direction: metered OAuth connections exist (cloud-billed OAuth), and
- * plan-included API-key connections exist (Copilot seat tokens). So this is a
- * CURATED table, hand-set from each provider's published terms — deliberately
- * the same pattern `FreeModelBudget.hardStopGuaranteed`
- * (`open-sse/config/freeModelCatalog.ts`) already established: a fact about
- * the upstream's commercial terms, never derived from `authType` and never
- * inferred from a live API response.
- *
- * Uncurated is not "free": anything absent here resolves to `unknown`, which
- * every consumer treats as `metered`. New providers therefore start OUTSIDE
- * the subscription rung and have to be curated in deliberately — the same
- * fail-safe direction STRICT_ZERO_COST uses for uncatalogued models.
- */
-
-/**
- * How a connection's upstream charges for the requests it serves.
- *
- * - `subscription` — covered by a flat-rate plan the operator already pays.
- *   Consuming its quota costs nothing incremental; the plan is a sunk cost.
- * - `metered` — pay-per-token / pay-per-credit. Every request adds spend.
- * - `keyless` — no credential exists at all, so no request can be billed by
- *   construction (the synthetic no-auth path).
- * - `unknown` — not curated. Consumed as `metered` everywhere.
- */
-export type ConnectionBillingClass = "subscription" | "metered" | "keyless" | "unknown";
-
-/**
- * What happens when a subscription plan's allowance runs out.
- *
- * - `hard-stop` — the upstream refuses further requests until the window
- *   resets. Exhaustion cannot cost money, so such a connection is admissible
- *   to the strictest "never spend a cent extra" grouping.
- * - `meters-to-paid` — the upstream keeps serving and bills the overage.
- *   Perfectly usable while quota remains, never admissible to the strict
- *   grouping.
- * - `unknown` — not established. Treated exactly like `meters-to-paid` by
- *   every consumer; it is the conservative default for a provider whose terms
- *   allow an operator to opt into usage-based billing past the plan.
- */
-export type ConnectionOverageBehavior = "hard-stop" | "meters-to-paid" | "unknown";
-
-export interface ConnectionBillingEntry {
-  /** Provider id as registered in `open-sse/config/providers/registry/`. */
-  provider: string;
-  /**
-   * Restricts the entry to connections whose `authType` matches. Omit for a
-   * provider-wide entry. A matching auth-typed entry always wins over the
-   * provider-wide one, so a provider offering both a plan-included OAuth login
-   * and a metered API key can declare both.
-   */
-  authType?: string;
-  billing: ConnectionBillingClass;
-  overage: ConnectionOverageBehavior;
-  /** Operator-visible justification for the classification. */
-  reason: string;
-}
-
-/**
- * Curated entries. Conservative by design — a provider whose terms let the
- * operator enable usage-based billing past the plan is recorded as `unknown`
- * overage, not `hard-stop`, because the strict grouping's entire promise is
- * that it cannot surprise you.
- *
- * Providers already classified free by the economic tier resolver (`kiro`,
- * `qoder`, and the rest of `LEGACY_FREE_PROVIDERS` /
- * `deriveNoAuthFreeProviders()` in `open-sse/services/tierConfig.ts`) are
- * deliberately NOT listed here: they land on the ladder's `free` rung through
- * `classifyTier()` and would only be double-claimed by an entry here.
- */
-export const CONNECTION_BILLING_CATALOG: readonly ConnectionBillingEntry[] = [
-  {
-    provider: "claude",
-    authType: "oauth",
-    billing: "subscription",
-    overage: "hard-stop",
-    reason:
-      "Claude Code OAuth serves the operator's Anthropic Pro/Max plan windows. " +
-      "Exceeding a window is refused until it resets; no per-token charge accrues.",
-  },
-  {
-    provider: "codex",
-    authType: "oauth",
-    billing: "subscription",
-    overage: "hard-stop",
-    reason:
-      "Codex OAuth serves the ChatGPT plan's included Codex quota. Exhaustion is " +
-      "refused until the plan window resets rather than billed.",
-  },
-  {
-    provider: "antigravity",
-    authType: "oauth",
-    billing: "subscription",
-    overage: "hard-stop",
-    reason:
-      "Antigravity OAuth serves built-in plan quotas that stop serving once consumed; " +
-      "OmniRoute already tracks their reset windows (see antigravityCredits.ts).",
-  },
-  {
-    provider: "cursor",
-    authType: "oauth",
-    billing: "subscription",
-    overage: "unknown",
-    reason:
-      "Cursor Pro includes a request allowance, but usage-based pricing past the plan " +
-      "can be enabled per account and OmniRoute cannot observe that setting. Recorded " +
-      "as unknown overage so the strict grouping excludes it.",
-  },
-  {
-    provider: "copilot-web",
-    authType: "apikey",
-    billing: "subscription",
-    overage: "unknown",
-    reason:
-      "GitHub Copilot is a per-seat subscription (the credential is a seat token, not a " +
-      "metered API key), but additional premium requests can be billed when the account " +
-      "opts in. Recorded as unknown overage.",
-  },
-  {
-    provider: "devin-desktop",
-    authType: "oauth",
-    billing: "subscription",
-    overage: "meters-to-paid",
-    reason:
-      "Devin Desktop draws on the plan's included ACUs and continues billing past them, " +
-      "so it is plan-included while quota remains but never overage-safe.",
-  },
-];

open-sse/config/embeddingRegistry.ts

@@ -408,6 +408,7 @@ export const EMBEDDING_PROVIDERS: Record<string, EmbeddingProvider> = {
       },
     ],
   },
+
 };
 
 const EMBEDDING_PROVIDER_ALIASES: Record<string, string> = {
@@ -470,6 +471,38 @@ export function getEmbeddingProvider(providerId: string): EmbeddingProvider | nu
   return EMBEDDING_PROVIDERS[resolveEmbeddingProviderId(providerId)] || null;
 }
 
+/**
+ * Derive an OpenAI-compatible embeddings config for a chat provider that has NO
+ * curated EMBEDDING_PROVIDERS entry. Works for any registry provider whose base
+ * URL ends in /chat/completions by swapping that suffix for /embeddings (groq,
+ * mistral, together, upstage, fireworks, nvidia, vercel-ai-gateway, ...).
+ * Dynamic-URL providers (no usable static base) derive to
+ * null — they need bespoke URL handling, not a bogus endpoint.
+ *
+ * This is a FALLBACK only: callers must check getEmbeddingProvider() first so
+ * curated entries keep their specialized configuration.
+ */
+export function deriveEmbeddingProviderForChatProvider(
+  providerId: string,
+  chatEntry: { id?: string; baseUrl?: string | string[] } | null | undefined
+): EmbeddingProvider | null {
+  if (!chatEntry) return null;
+  const rawBase = Array.isArray(chatEntry.baseUrl)
+    ? chatEntry.baseUrl[0]
+    : chatEntry.baseUrl;
+  if (!rawBase || typeof rawBase !== "string") return null;
+  // stripTrailingSlashes-equivalent without importing open-sse utils here:
+  const base = rawBase.replace(/\/+$/, "");
+  if (!base.endsWith("/chat/completions")) return null;
+  return {
+    id: providerId,
+    baseUrl: `${base.slice(0, -"/chat/completions".length)}/embeddings`,
+    authType: "apikey",
+    authHeader: "bearer",
+    models: [],
+  };
+}
+
 /**
  * Parse embedding model string (format: "provider/model" or just "model")
  * Returns { provider, model }
@@ -485,6 +518,18 @@ export function parseEmbeddingModel(
   const slashIdx = modelStr.indexOf("/");
   if (slashIdx > 0) {
     const rawProvider = modelStr.slice(0, slashIdx);
+
+    // A configured provider_node whose prefix exactly equals the requested
+    // provider segment always wins — even when that segment is also an alias
+    // of a curated provider (a local node must not be hijacked by a registry
+    // alias). Same exact-match precedence documented for
+    // EMBEDDING_MODEL_ALIASES above.
+    const dynamicExact =
+      dynamicProviders && dynamicProviders.find((dp) => dp.id === rawProvider);
+    if (dynamicExact) {
+      return { provider: rawProvider, model: modelStr.slice(slashIdx + 1) };
+    }
+
     const resolvedProvider = resolveEmbeddingProviderId(rawProvider);
 
     if (EMBEDDING_PROVIDERS[resolvedProvider]) {

open-sse/config/opencodeZenGoSharedModels.ts

@@ -12,5 +12,15 @@
 export const OPENCODE_ZEN_GO_SHARED_MODELS = Object.freeze([
   { id: "kimi-k2.7-code", name: "Kimi K2.7 Code" },
   { id: "qwen3.5-plus", name: "Qwen3.5 Plus", targetFormat: "claude", supportsVision: false },
-  { id: "qwen3.6-plus", name: "Qwen3.6 Plus", targetFormat: "claude", supportsVision: false },
+  {
+    id: "qwen3.6-plus",
+    name: "Qwen3.6 Plus",
+    targetFormat: "claude",
+    supportsVision: false,
+    // #10788: effort-tier aliases exist as explicit registry rows; declare the
+    // vocabulary on the shared base row so variant synthesis and the sanitizer
+    // agree on it for both opencode-go and opencode-zen.
+    supportsReasoning: true,
+    supportedThinkingEfforts: ["high", "max"],
+  },
 ]);

open-sse/config/providers/registry/nvidia/index.ts

@@ -18,7 +18,17 @@ export const nvidiaProvider: RegistryEntry = {
   passthroughModels: true,
   models: [
     // #6108: z-ai/glm-5.1 EOL'd 2026-07-02 (direct probe returns 410) — dropped.
-    { id: "z-ai/glm-5.2", name: "GLM 5.2" },
+    // #10788: NVIDIA's hosted GLM-5.2 exposes a BINARY thinking switch
+    // (chat_template_kwargs.enable_thinking), not effort tiers — see
+    // mapNvidiaGlm52ReasoningParams. Declaring an empty tier list keeps the
+    // catalog from synthesizing unresolvable -low/-high/-max variant ids while
+    // still marking the model reasoning-capable.
+    {
+      id: "z-ai/glm-5.2",
+      name: "GLM 5.2",
+      supportsReasoning: true,
+      supportedThinkingEfforts: [],
+    },
     // #3329/#6108: minimaxai/minimax-m3 stays excluded from the nvidia tier — it
     // still 404s here for most callers; the single 200 probe in #6108 was not
     // reproducible enough to override the #3329 guard. Re-add only once NVIDIA

open-sse/config/providers/registry/opencode/go/index.ts

@@ -19,9 +19,15 @@ export const opencode_goProvider: RegistryEntry = {
     // `kimi-k2.7-code` (the live API rejects the plain `kimi-k2.7` alias for
     // `/chat/completions`, even though the docs config example uses it).
     // GLM-5.2 — base model + effort-tier aliases (#6922).
-    // OpencodeExecutor rewrites the alias to the canonical id and injects
-    // reasoning_effort, mirroring the deepseek-v4-pro-* pattern.
-    { id: "glm-5.2", name: "GLM-5.2", supportsReasoning: true },
+    // #10788: the tier vocabulary is declared on the base row so the catalog's
+    // variant synthesis (#9485) and the effort sanitizer share one source of
+    // truth with OpencodeExecutor's EFFORT_TIERS.
+    {
+      id: "glm-5.2",
+      name: "GLM-5.2",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
     { id: "glm-5.2-high", name: "GLM-5.2 (high effort)", supportsReasoning: true },
     { id: "glm-5.2-max", name: "GLM-5.2 (max effort)", supportsReasoning: true },
 
@@ -34,11 +40,16 @@ export const opencode_goProvider: RegistryEntry = {
     { id: "kimi-k2.6", name: "Kimi K2.6" },
     { id: "kimi-k2.5", name: "Kimi K2.5" },
     // #8353: Kimi K3 base + max-effort alias from the OpenCode Go registry.
-    { id: "kimi-k3", name: "Kimi K3", supportsReasoning: true },
+    { id: "kimi-k3", name: "Kimi K3", supportsReasoning: true, supportedThinkingEfforts: ["max"] },
     { id: "kimi-k3-max", name: "Kimi K3 (max effort)", supportsReasoning: true },
     // MiMo-V2.5 — base model + effort-tier aliases (#6922).
     { id: "mimo-v2.5-pro", name: "MiMo-V2.5-Pro", supportsReasoning: true },
-    { id: "mimo-v2.5", name: "MiMo-V2.5", supportsReasoning: true },
+    {
+      id: "mimo-v2.5",
+      name: "MiMo-V2.5",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
     { id: "mimo-v2.5-high", name: "MiMo-V2.5 (high effort)", supportsReasoning: true },
     { id: "mimo-v2.5-max", name: "MiMo-V2.5 (max effort)", supportsReasoning: true },
     // #3110: MiniMax M3 via OpenCode Go tier
@@ -59,7 +70,14 @@ export const opencode_goProvider: RegistryEntry = {
     // so combo routing skips them when the request contains image blocks,
     // preventing image content from reaching a vision-incapable upstream.
     // #8353: effort-tier aliases from the OpenCode Go registry.
-    { id: "qwen3.7-max", name: "Qwen3.7 Max", targetFormat: "claude", supportsVision: false },
+    {
+      id: "qwen3.7-max",
+      name: "Qwen3.7 Max",
+      targetFormat: "claude",
+      supportsVision: false,
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
     {
       id: "qwen3.7-max-high",
       name: "Qwen3.7 Max (high effort)",
@@ -79,6 +97,8 @@ export const opencode_goProvider: RegistryEntry = {
       name: "Qwen3.7 Plus",
       targetFormat: "claude",
       supportsVision: false,
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
     },
     {
       id: "qwen3.7-plus-high",
@@ -111,7 +131,13 @@ export const opencode_goProvider: RegistryEntry = {
       supportsReasoning: true,
     },
     // #8353: hy3 is the Go-tier base id (distinct from hy3-preview / hy3-free).
-    { id: "hy3", name: "Hunyuan3", contextLength: 256000, supportsReasoning: true },
+    {
+      id: "hy3",
+      name: "Hunyuan3",
+      contextLength: 256000,
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["none", "low", "high"],
+    },
     {
       id: "hy3-none",
       name: "Hunyuan3 (none effort)",
@@ -201,7 +227,12 @@ export const opencode_goProvider: RegistryEntry = {
       targetFormat: "openai-responses",
     },
     // #8353: Grok 4.5 + effort tiers from the OpenCode Go registry.
-    { id: "grok-4.5", name: "Grok 4.5", supportsReasoning: true },
+    {
+      id: "grok-4.5",
+      name: "Grok 4.5",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["low", "medium", "high"],
+    },
     { id: "grok-4.5-low", name: "Grok 4.5 (low effort)", supportsReasoning: true },
     { id: "grok-4.5-medium", name: "Grok 4.5 (medium effort)", supportsReasoning: true },
     { id: "grok-4.5-high", name: "Grok 4.5 (high effort)", supportsReasoning: true },

open-sse/config/providers/registry/opencode/zen/index.ts

@@ -79,18 +79,35 @@ export const opencode_zenProvider: RegistryEntry = {
     },
 
     // ── DeepSeek ────────────────────────────────────────────────
-    { id: "deepseek-v4-pro", name: "DeepSeek V4 Pro" },
-    { id: "deepseek-v4-flash", name: "DeepSeek V4 Flash" },
+    // #10788: same tier vocabulary as opencode-go's DeepSeek rows — the Zen
+    // upstream accepts the identical effort set on these models.
+    {
+      id: "deepseek-v4-pro",
+      name: "DeepSeek V4 Pro",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["none", "low", "high", "max"],
+    },
+    {
+      id: "deepseek-v4-flash",
+      name: "DeepSeek V4 Flash",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["none", "low", "high", "max"],
+    },
 
     // ── GLM / Z.AI ─────────────────────────────────────────────
-    { id: "glm-5.2", name: "GLM-5.2" },
+    {
+      id: "glm-5.2",
+      name: "GLM-5.2",
+      supportsReasoning: true,
+      supportedThinkingEfforts: ["high", "max"],
+    },
 
     // ── MiniMax ────────────────────────────────────────────────
     // #3110: MiniMax M3 — frontier coding model with 1M context
     { id: "minimax-m3", name: "MiniMax M3", contextLength: 1048576, supportsVision: true },
 
     // ── Kimi / Moonshot ────────────────────────────────────────
-    { id: "kimi-k3", name: "Kimi K3" },
+    { id: "kimi-k3", name: "Kimi K3", supportsReasoning: true, supportedThinkingEfforts: ["max"] },
     // kimi-k2.7-code declared identically on opencode-go — see OPENCODE_ZEN_GO_SHARED_MODELS.
 
     // ── Qwen ───────────────────────────────────────────────────

open-sse/config/rerankRegistry.ts

@@ -218,3 +218,29 @@ export function getAllRerankModels() {
   }
   return models;
 }
+
+/**
+ * Derive a Cohere-compatible rerank config for a chat provider that has NO
+ * curated RERANK_PROVIDERS entry. Works for any registry provider whose base
+ * URL ends in /chat/completions by swapping that suffix for /rerank (groq,
+ * mistral, vercel-ai-gateway, ...). Dynamic-URL providers (no usable static
+ * base, e.g. dynamic account-scoped hosts) derive to null — they need bespoke
+ * URL handling.
+ *
+ * This is a FALLBACK only: callers must check getRerankProvider() first so
+ * curated entries keep their specialized configuration and format adapters.
+ */
+export function deriveRerankProviderForChatProvider(providerId, chatEntry) {
+  if (!chatEntry) return null;
+  const rawBase = Array.isArray(chatEntry.baseUrl) ? chatEntry.baseUrl[0] : chatEntry.baseUrl;
+  if (!rawBase || typeof rawBase !== "string") return null;
+  const base = rawBase.replace(/\/+$/, "");
+  if (!base.endsWith("/chat/completions")) return null;
+  return {
+    id: providerId,
+    baseUrl: `${base.slice(0, -"/chat/completions".length)}/rerank`,
+    authType: "apikey",
+    authHeader: "bearer",
+    models: [],
+  };
+}

open-sse/executors/index.ts

@@ -1,246 +1,218 @@
 import { SEARCH_PROVIDERS } from "../config/searchRegistry.ts";
-import { registerExecutor, getRegisteredExecutor, hasRegisteredExecutor } from "./registry.ts";
+import {
+  registerLazyExecutor,
+  loadRegisteredExecutor,
+  hasRegisteredExecutor,
+} from "./registry.ts";
+// Type-only: pulls no runtime code, keeps DefaultExecutor the only eager class.
 import type { BaseExecutor } from "./base.ts";
-import { AntigravityExecutor } from "./antigravity.ts";
-import { GithubExecutor } from "./github.ts";
-import { GheCopilotExecutor } from "./ghe-copilot.ts";
-import { QoderExecutor } from "./qoder.ts";
-import { KiroExecutor } from "./kiro.ts";
-import { CodexExecutor } from "./codex.ts";
-import { CodexAppServerExecutor } from "./codex-app-server.ts";
-import { CursorExecutor } from "./cursor.ts";
-import { TraeExecutor } from "./trae.ts";
 import { DefaultExecutor } from "./default.ts";
-import { BedrockExecutor } from "./bedrock.ts";
-import { GlmExecutor } from "./glm.ts";
-import { PollinationsExecutor } from "./pollinations.ts";
-import { CloudflareAIExecutor } from "./cloudflare-ai.ts";
-import { FreebuffExecutor } from "./freebuff.ts";
-import { OpencodeExecutor } from "./opencode.ts";
-import { VertexExecutor } from "./vertex.ts";
-import { CliproxyapiExecutor } from "./cliproxyapi.ts";
-import { DarioExecutor } from "./dario.ts";
-import { NineRouterExecutor } from "./ninerouter.ts";
-import { PerplexityWebExecutor } from "./perplexity-web.ts";
-import { GrokWebExecutor } from "./grok-web.ts";
-import { GeminiWebExecutor } from "./gemini-web.ts";
-import { TencentAIStudioWebExecutor } from "./tencent-aistudio-web.ts";
-import { GeminiBusinessExecutor } from "./gemini-business.ts";
-import { ChatGptWebExecutor } from "./chatgpt-web.ts";
-import { ChatGptWebCodexExecutor } from "./chatgpt-web-codex.ts";
-import { BlackboxWebExecutor } from "./blackbox-web.ts";
-import { MuseSparkWebExecutor } from "./muse-spark-web.ts";
-import { AzureOpenAIExecutor } from "./azure-openai.ts";
-import { AzureAiExecutor } from "./azure-ai.ts";
-import { CommandCodeExecutor } from "./commandCode.ts";
-import { GitlabExecutor } from "./gitlab.ts";
-import { NlpCloudExecutor } from "./nlpcloud.ts";
-import { DevinDesktopExecutor } from "./devin-desktop.ts";
-import { ZedHostedExecutor } from "./zed-hosted.ts";
-import { DevinCliExecutor } from "./devin-cli.ts";
-import { ZcodeExecutor } from "./zcode.ts";
-import { DevinCliAgenticExecutor } from "./devin-cli-agentic.ts";
-import { AuggieExecutor } from "./auggie.ts";
-import { DeepSeekWebExecutor } from "./deepseek-web.ts";
-import { DeepSeekWebWithAutoRefreshExecutor } from "./deepseek-web-with-auto-refresh.ts";
-import { AdaptaWebExecutor } from "./adapta-web.ts";
-import { CopilotWebExecutor } from "./copilot-web.ts";
-import { CopilotM365WebExecutor } from "./copilot-m365-web.ts";
-import { MicrosoftDesignerWebExecutor } from "./microsoft-designer-web.ts";
-import { AdobeFireflyExecutor } from "./adobe-firefly.ts";
-import { VeoAIFreeWebExecutor } from "./veoaifree-web.ts";
-import { DuckDuckGoWebExecutor } from "./duckduckgo-web.ts";
-import { FeloWebExecutor } from "./felo-web.ts";
-import { T3ChatWebExecutor } from "./t3-chat-web.ts";
-import { ClaudeWebExecutor } from "./claude-web.ts";
-import { InnerAiExecutor } from "./inner-ai.ts";
-import { HuggingChatExecutor } from "./huggingchat.ts";
-import { YuanbaoWebExecutor } from "./yuanbao-web.ts";
-import { PoeWebExecutor } from "./poe-web.ts";
-import { VeniceWebExecutor } from "./venice-web.ts";
-import { NotionWebExecutor } from "./notion-web.ts";
-import { V0VercelWebExecutor } from "./v0-vercel-web.ts";
-import { CheaperInferenceExecutor } from "./cheaperinference.ts";
-import { KimiWebExecutor } from "./kimi-web.ts";
-import { DoubaoWebExecutor } from "./doubao-web.ts";
-import { QwenWebExecutor } from "./qwen-web.ts";
-import { RaycastExecutor } from "./raycast.ts";
-import { HailuoWebExecutor } from "./hailuo-web.ts";
-import { ZaiWebExecutor } from "./zai-web.ts";
-import { KimiExecutor } from "./kimi.ts";
-import { MoonshotExecutor } from "./moonshot.ts";
-import { TheOldLlmExecutor } from "./theoldllm.ts";
-import { ChipotleExecutor } from "./chipotle.ts";
-import { LMArenaExecutor } from "./lmarena.ts";
-import { GrokCliExecutor } from "./grok-cli.ts";
-import { CodeBuddyCnExecutor } from "./codebuddy-cn.ts";
-import { ZenmuxFreeExecutor } from "./zenmux-free.ts";
-import { CloudflarePlaygroundExecutor } from "./cloudflare-playground.ts";
-import { TinyCmsExecutor } from "./tinycms.ts";
-import { HyperAgentExecutor } from "./hyperagent.ts";
-import { XaiExecutor } from "./xai.ts";
-import { PromptQlExecutor } from "./promptql.ts";
-import { ConolWebExecutor } from "./conol-web.ts";
 
-// R0.3 — declarative built-in table. The object literal stays as the single
-// place built-ins are declared (compile-time duplicate-key safety; the
-// check:known-symbols gate parses this literal from source), but lookup goes
-// through the ExecutorRegistry (./registry.ts): every entry is registered at
-// module load below, and getExecutor()/hasSpecializedExecutor() consult the
-// registry — the literal is never read at request time.
-const executors = {
-  antigravity: new AntigravityExecutor(),
-  agy: new AntigravityExecutor(),
-  github: new GithubExecutor(),
-  "ghe-copilot": new GheCopilotExecutor(),
-  qoder: new QoderExecutor(),
-  kiro: new KiroExecutor(),
-  "amazon-q": new KiroExecutor("amazon-q"),
-  bedrock: new BedrockExecutor(),
-  codex: new CodexExecutor(),
-  "codex-app-server": new CodexAppServerExecutor({}, "codex-app-server"),
-  "chatgpt-web-codex": new ChatGptWebCodexExecutor(),
-  "cgpt-codex": new ChatGptWebCodexExecutor(),
-  cursor: new CursorExecutor(),
-  trae: new TraeExecutor(),
-  glm: new GlmExecutor("glm"),
-  "glm-cn": new GlmExecutor("glm-cn"),
-  glmt: new GlmExecutor("glmt"),
-  cu: new CursorExecutor(), // Alias for cursor
-  "cursor-api": new CursorExecutor("cursor-api"),
-  cua: new CursorExecutor("cursor-api"),
-  "azure-openai": new AzureOpenAIExecutor(),
-  "azure-ai": new AzureAiExecutor(),
-  "command-code": new CommandCodeExecutor(),
-  cmd: new CommandCodeExecutor(), // Alias
-  gitlab: new GitlabExecutor(),
-  "gitlab-duo": new GitlabExecutor("gitlab-duo"),
-  nlpcloud: new NlpCloudExecutor(),
-  pollinations: new PollinationsExecutor(),
-  pol: new PollinationsExecutor(), // Alias
-  "cloudflare-ai": new CloudflareAIExecutor(),
-  cf: new CloudflareAIExecutor(), // Alias
-  freebuff: new FreebuffExecutor(),
-  fb: new FreebuffExecutor(), // Alias
-  "opencode-zen": new OpencodeExecutor("opencode-zen"),
-  "opencode-go": new OpencodeExecutor("opencode-go"),
-  opencode: new OpencodeExecutor("opencode-zen"), // Alias for opencode-zen
-  vertex: new VertexExecutor(),
-  "vertex-partner": new VertexExecutor(),
-  cliproxyapi: new CliproxyapiExecutor(),
-  cpa: new CliproxyapiExecutor(), // Alias
-  dario: new DarioExecutor(),
-  dr: new DarioExecutor(), // Alias
-  "9router": new NineRouterExecutor(),
-  nr: new NineRouterExecutor(), // Alias
-  "perplexity-web": new PerplexityWebExecutor(),
-  "pplx-web": new PerplexityWebExecutor(), // Alias
-  "grok-web": new GrokWebExecutor(),
-  "claude-web": new ClaudeWebExecutor(),
-  "cw-web": new ClaudeWebExecutor(), // Alias
-  "gemini-web": new GeminiWebExecutor(),
-  gweb: new GeminiWebExecutor(), // Alias
-  "gemini-business": new GeminiBusinessExecutor(),
-  gembiz: new GeminiBusinessExecutor(), // Alias
-  "chatgpt-web": new ChatGptWebExecutor(),
-  "cgpt-web": new ChatGptWebExecutor(), // Alias
-  "blackbox-web": new BlackboxWebExecutor(),
-  "bb-web": new BlackboxWebExecutor(), // Alias
-  "muse-spark-web": new MuseSparkWebExecutor(),
-  "ms-web": new MuseSparkWebExecutor(), // Alias
-  "devin-desktop": new DevinDesktopExecutor(),
-  "zed-hosted": new ZedHostedExecutor(),
-  "devin-cli": new DevinCliExecutor(),
-  zcode: new ZcodeExecutor(),
-  zc: new ZcodeExecutor(), // Alias
-  "devin-cli-agentic": new DevinCliAgenticExecutor(),
-  devin: new DevinCliExecutor(), // Alias
-  "deepseek-web": new DeepSeekWebWithAutoRefreshExecutor(),
-  "ds-web": new DeepSeekWebWithAutoRefreshExecutor(), // Alias
-  "adapta-web": new AdaptaWebExecutor(),
-  "adp-web": new AdaptaWebExecutor(), // Alias
-  "copilot-web": new CopilotWebExecutor(),
-  "copilot-m365-web": new CopilotM365WebExecutor(),
-  copilot: new CopilotWebExecutor(), // Alias
-  "microsoft-designer-web": new MicrosoftDesignerWebExecutor(),
-  msdesigner: new MicrosoftDesignerWebExecutor(), // Alias
-  "adobe-firefly": new AdobeFireflyExecutor(),
-  firefly: new AdobeFireflyExecutor(), // Alias
-  "veoaifree-web": new VeoAIFreeWebExecutor(),
-  "veo-free": new VeoAIFreeWebExecutor(), // Alias
-  "duckduckgo-web": new DuckDuckGoWebExecutor(),
-  ddgw: new DuckDuckGoWebExecutor(), // Alias
-  "felo-web": new FeloWebExecutor(),
-  felo: new FeloWebExecutor(), // Alias
-  "t3-web": new T3ChatWebExecutor(),
-  t3chat: new T3ChatWebExecutor(), // Alias
-  "inner-ai": new InnerAiExecutor(),
-  "in-ai": new InnerAiExecutor(), // Alias
-  huggingchat: new HuggingChatExecutor(),
-  hc: new HuggingChatExecutor(), // Alias
-  "yuanbao-web": new YuanbaoWebExecutor(),
-  "tencent-aistudio-web": new TencentAIStudioWebExecutor(),
-  tasw: new TencentAIStudioWebExecutor(),
-  ybw: new YuanbaoWebExecutor(), // Alias
-  "poe-web": new PoeWebExecutor(),
+// R0.3 — declarative built-in table, made LAZY by #11220.
+//
+// The object literal below stays as the single place built-ins are declared
+// (compile-time duplicate-key safety; the check:known-symbols gate parses this
+// literal from source), but its values are now deferred loaders instead of
+// eagerly-constructed instances. At module load every ALIAS is registered in
+// declaration order; the class import + construction happen on first use via
+// loadRegisteredExecutor() and are cached in the same registry a static
+// registration would have populated.
+//
+// Why: importing this barrel previously pulled all ~100 executor modules and
+// constructed every instance at boot — ~0.7–1.2s and ~35MB of heap measured on
+// top of the minimal set — even for deployments that use a handful of
+// providers. Bundlers split the dynamic imports into on-demand chunks.
+//
+// Contract preserved (pinned by tests/unit/executor-map-golden.test.ts):
+//   - keys and their ORDER are byte-stable
+//   - each alias still gets its OWN instance (aliases never share)
+//   - ctor arguments are unchanged
+const lazyExecutors: Record<string, () => Promise<BaseExecutor>> = {
+  antigravity: () => import("./antigravity.ts").then((m) => new m.AntigravityExecutor()),
+  agy: () => import("./antigravity.ts").then((m) => new m.AntigravityExecutor()),
+  github: () => import("./github.ts").then((m) => new m.GithubExecutor()),
+  "ghe-copilot": () => import("./ghe-copilot.ts").then((m) => new m.GheCopilotExecutor()),
+  qoder: () => import("./qoder.ts").then((m) => new m.QoderExecutor()),
+  kiro: () => import("./kiro.ts").then((m) => new m.KiroExecutor()),
+  "amazon-q": () => import("./kiro.ts").then((m) => new m.KiroExecutor("amazon-q")),
+  bedrock: () => import("./bedrock.ts").then((m) => new m.BedrockExecutor()),
+  codex: () => import("./codex.ts").then((m) => new m.CodexExecutor()),
+  "codex-app-server": () =>
+    import("./codex-app-server.ts").then(
+      (m) => new m.CodexAppServerExecutor({}, "codex-app-server")
+    ),
+  "chatgpt-web-codex": () =>
+    import("./chatgpt-web-codex.ts").then((m) => new m.ChatGptWebCodexExecutor()),
+  "cgpt-codex": () =>
+    import("./chatgpt-web-codex.ts").then((m) => new m.ChatGptWebCodexExecutor()),
+  cursor: () => import("./cursor.ts").then((m) => new m.CursorExecutor()),
+  trae: () => import("./trae.ts").then((m) => new m.TraeExecutor()),
+  glm: () => import("./glm.ts").then((m) => new m.GlmExecutor("glm")),
+  "glm-cn": () => import("./glm.ts").then((m) => new m.GlmExecutor("glm-cn")),
+  glmt: () => import("./glm.ts").then((m) => new m.GlmExecutor("glmt")),
+  cu: () => import("./cursor.ts").then((m) => new m.CursorExecutor()), // Alias for cursor
+  "cursor-api": () => import("./cursor.ts").then((m) => new m.CursorExecutor("cursor-api")),
+  cua: () => import("./cursor.ts").then((m) => new m.CursorExecutor("cursor-api")),
+  "azure-openai": () => import("./azure-openai.ts").then((m) => new m.AzureOpenAIExecutor()),
+  "azure-ai": () => import("./azure-ai.ts").then((m) => new m.AzureAiExecutor()),
+  "command-code": () => import("./commandCode.ts").then((m) => new m.CommandCodeExecutor()),
+  cmd: () => import("./commandCode.ts").then((m) => new m.CommandCodeExecutor()), // Alias
+  gitlab: () => import("./gitlab.ts").then((m) => new m.GitlabExecutor()),
+  "gitlab-duo": () => import("./gitlab.ts").then((m) => new m.GitlabExecutor("gitlab-duo")),
+  nlpcloud: () => import("./nlpcloud.ts").then((m) => new m.NlpCloudExecutor()),
+  pollinations: () => import("./pollinations.ts").then((m) => new m.PollinationsExecutor()),
+  pol: () => import("./pollinations.ts").then((m) => new m.PollinationsExecutor()), // Alias
+  "cloudflare-ai": () => import("./cloudflare-ai.ts").then((m) => new m.CloudflareAIExecutor()),
+  cf: () => import("./cloudflare-ai.ts").then((m) => new m.CloudflareAIExecutor()), // Alias
+  freebuff: () => import("./freebuff.ts").then((m) => new m.FreebuffExecutor()),
+  fb: () => import("./freebuff.ts").then((m) => new m.FreebuffExecutor()), // Alias
+  "opencode-zen": () =>
+    import("./opencode.ts").then((m) => new m.OpencodeExecutor("opencode-zen")),
+  "opencode-go": () =>
+    import("./opencode.ts").then((m) => new m.OpencodeExecutor("opencode-go")),
+  opencode: () =>
+    import("./opencode.ts").then((m) => new m.OpencodeExecutor("opencode-zen")), // Alias for opencode-zen
+  vertex: () => import("./vertex.ts").then((m) => new m.VertexExecutor()),
+  "vertex-partner": () => import("./vertex.ts").then((m) => new m.VertexExecutor()),
+  cliproxyapi: () => import("./cliproxyapi.ts").then((m) => new m.CliproxyapiExecutor()),
+  cpa: () => import("./cliproxyapi.ts").then((m) => new m.CliproxyapiExecutor()), // Alias
+  dario: () => import("./dario.ts").then((m) => new m.DarioExecutor()),
+  dr: () => import("./dario.ts").then((m) => new m.DarioExecutor()), // Alias
+  "9router": () => import("./ninerouter.ts").then((m) => new m.NineRouterExecutor()),
+  nr: () => import("./ninerouter.ts").then((m) => new m.NineRouterExecutor()), // Alias
+  "perplexity-web": () =>
+    import("./perplexity-web.ts").then((m) => new m.PerplexityWebExecutor()),
+  "pplx-web": () =>
+    import("./perplexity-web.ts").then((m) => new m.PerplexityWebExecutor()), // Alias
+  "grok-web": () => import("./grok-web.ts").then((m) => new m.GrokWebExecutor()),
+  "claude-web": () => import("./claude-web.ts").then((m) => new m.ClaudeWebExecutor()),
+  "cw-web": () => import("./claude-web.ts").then((m) => new m.ClaudeWebExecutor()), // Alias
+  "gemini-web": () => import("./gemini-web.ts").then((m) => new m.GeminiWebExecutor()),
+  gweb: () => import("./gemini-web.ts").then((m) => new m.GeminiWebExecutor()), // Alias
+  "gemini-business": () =>
+    import("./gemini-business.ts").then((m) => new m.GeminiBusinessExecutor()),
+  gembiz: () =>
+    import("./gemini-business.ts").then((m) => new m.GeminiBusinessExecutor()), // Alias
+  "chatgpt-web": () => import("./chatgpt-web.ts").then((m) => new m.ChatGptWebExecutor()),
+  "cgpt-web": () => import("./chatgpt-web.ts").then((m) => new m.ChatGptWebExecutor()), // Alias
+  "blackbox-web": () => import("./blackbox-web.ts").then((m) => new m.BlackboxWebExecutor()),
+  "bb-web": () => import("./blackbox-web.ts").then((m) => new m.BlackboxWebExecutor()), // Alias
+  "muse-spark-web": () =>
+    import("./muse-spark-web.ts").then((m) => new m.MuseSparkWebExecutor()),
+  "ms-web": () => import("./muse-spark-web.ts").then((m) => new m.MuseSparkWebExecutor()), // Alias
+  "devin-desktop": () => import("./devin-desktop.ts").then((m) => new m.DevinDesktopExecutor()),
+  "zed-hosted": () => import("./zed-hosted.ts").then((m) => new m.ZedHostedExecutor()),
+  "devin-cli": () => import("./devin-cli.ts").then((m) => new m.DevinCliExecutor()),
+  zcode: () => import("./zcode.ts").then((m) => new m.ZcodeExecutor()),
+  zc: () => import("./zcode.ts").then((m) => new m.ZcodeExecutor()), // Alias
+  "devin-cli-agentic": () =>
+    import("./devin-cli-agentic.ts").then((m) => new m.DevinCliAgenticExecutor()),
+  devin: () => import("./devin-cli.ts").then((m) => new m.DevinCliExecutor()), // Alias
+  "deepseek-web": () =>
+    import("./deepseek-web-with-auto-refresh.ts").then(
+      (m) => new m.DeepSeekWebWithAutoRefreshExecutor()
+    ),
+  "ds-web": () =>
+    import("./deepseek-web-with-auto-refresh.ts").then(
+      (m) => new m.DeepSeekWebWithAutoRefreshExecutor()
+    ), // Alias
+  "adapta-web": () => import("./adapta-web.ts").then((m) => new m.AdaptaWebExecutor()),
+  "adp-web": () => import("./adapta-web.ts").then((m) => new m.AdaptaWebExecutor()), // Alias
+  "copilot-web": () => import("./copilot-web.ts").then((m) => new m.CopilotWebExecutor()),
+  "copilot-m365-web": () =>
+    import("./copilot-m365-web.ts").then((m) => new m.CopilotM365WebExecutor()),
+  copilot: () => import("./copilot-web.ts").then((m) => new m.CopilotWebExecutor()), // Alias
+  "microsoft-designer-web": () =>
+    import("./microsoft-designer-web.ts").then((m) => new m.MicrosoftDesignerWebExecutor()),
+  msdesigner: () =>
+    import("./microsoft-designer-web.ts").then((m) => new m.MicrosoftDesignerWebExecutor()), // Alias
+  "adobe-firefly": () => import("./adobe-firefly.ts").then((m) => new m.AdobeFireflyExecutor()),
+  firefly: () => import("./adobe-firefly.ts").then((m) => new m.AdobeFireflyExecutor()), // Alias
+  "veoaifree-web": () => import("./veoaifree-web.ts").then((m) => new m.VeoAIFreeWebExecutor()),
+  "veo-free": () => import("./veoaifree-web.ts").then((m) => new m.VeoAIFreeWebExecutor()), // Alias
+  "duckduckgo-web": () =>
+    import("./duckduckgo-web.ts").then((m) => new m.DuckDuckGoWebExecutor()),
+  ddgw: () => import("./duckduckgo-web.ts").then((m) => new m.DuckDuckGoWebExecutor()), // Alias
+  "felo-web": () => import("./felo-web.ts").then((m) => new m.FeloWebExecutor()),
+  felo: () => import("./felo-web.ts").then((m) => new m.FeloWebExecutor()), // Alias
+  "t3-web": () => import("./t3-chat-web.ts").then((m) => new m.T3ChatWebExecutor()),
+  t3chat: () => import("./t3-chat-web.ts").then((m) => new m.T3ChatWebExecutor()), // Alias
+  "inner-ai": () => import("./inner-ai.ts").then((m) => new m.InnerAiExecutor()),
+  "in-ai": () => import("./inner-ai.ts").then((m) => new m.InnerAiExecutor()), // Alias
+  huggingchat: () => import("./huggingchat.ts").then((m) => new m.HuggingChatExecutor()),
+  hc: () => import("./huggingchat.ts").then((m) => new m.HuggingChatExecutor()), // Alias
+  "yuanbao-web": () => import("./yuanbao-web.ts").then((m) => new m.YuanbaoWebExecutor()),
+  "tencent-aistudio-web": () =>
+    import("./tencent-aistudio-web.ts").then((m) => new m.TencentAIStudioWebExecutor()),
+  tasw: () =>
+    import("./tencent-aistudio-web.ts").then((m) => new m.TencentAIStudioWebExecutor()), // Alias
+  ybw: () => import("./yuanbao-web.ts").then((m) => new m.YuanbaoWebExecutor()), // Alias
+  "poe-web": () => import("./poe-web.ts").then((m) => new m.PoeWebExecutor()),
   // #8969: do NOT alias canonical `poe` (API-key / api.poe.com) to PoeWebExecutor.
   // Registry declares executor:"default"; the hard-coded map previously won and
   // routed API-key traffic to GraphQL /api/gql_POST → HTTP 405.
-  "venice-web": new VeniceWebExecutor(),
-  ven: new VeniceWebExecutor(), // Alias
-  "notion-web": new NotionWebExecutor(),
-  nw: new NotionWebExecutor(), // Alias
-  promptql: new PromptQlExecutor(),
-  pql: new PromptQlExecutor(), // Alias
-  "v0-vercel-web": new V0VercelWebExecutor(),
-  v0: new V0VercelWebExecutor(), // Alias
-  "kimi-web": new KimiWebExecutor(),
-  "kimi-coding-apikey": new KimiExecutor("kimi-coding-apikey"), // Legacy alias
-  "kimi-coding": new KimiExecutor(), // Alias
-  moonshot: new MoonshotExecutor(),
-  kimi: new MoonshotExecutor("kimi"), // Hidden legacy Moonshot provider id
-  cheaperinference: new CheaperInferenceExecutor(),
-  cinf: new CheaperInferenceExecutor("cheaperinference"), // Alias
-  "doubao-web": new DoubaoWebExecutor(),
-  db: new DoubaoWebExecutor(), // Alias
-  "qwen-web": new QwenWebExecutor(),
-  raycast: new RaycastExecutor(),
-  rc: new RaycastExecutor(), // Alias
-  "hailuo-web": new HailuoWebExecutor(),
-  "zai-web": new ZaiWebExecutor(),
-  zw: new ZaiWebExecutor(), // Alias
-  theoldllm: new TheOldLlmExecutor(),
-  tllm: new TheOldLlmExecutor(), // Alias
-  chipotle: new ChipotleExecutor(),
-  pepper: new ChipotleExecutor(), // Alias
-  lmarena: new LMArenaExecutor(),
-  lma: new LMArenaExecutor(), // Alias
-  "grok-cli": new GrokCliExecutor(),
-  gc: new GrokCliExecutor(), // Alias
-  "codebuddy-cn": new CodeBuddyCnExecutor(),
-  cbcn: new CodeBuddyCnExecutor(), // Alias for codebuddy-cn
-  "zenmux-free": new ZenmuxFreeExecutor(),
-  "cloudflare-playground": new CloudflarePlaygroundExecutor(),
-  cfp: new CloudflarePlaygroundExecutor(), // Alias for cloudflare-playground
-  "tinycms-web": new TinyCmsExecutor(),
-  tcw: new TinyCmsExecutor(), // Alias
-  hyperagent: new HyperAgentExecutor(),
-  ha: new HyperAgentExecutor(), // Alias
-  zmf: new ZenmuxFreeExecutor(), // Alias for zenmux-free
-  auggie: new AuggieExecutor(),
-  xai: new XaiExecutor(),
-  "xai-oauth": new XaiExecutor("xai-oauth"),
-  xao: new XaiExecutor("xai-oauth"),
-  qw: new QwenWebExecutor(), // Alias
-  "conol-web": new ConolWebExecutor(),
-  cnl: new ConolWebExecutor(), // Alias
+  "venice-web": () => import("./venice-web.ts").then((m) => new m.VeniceWebExecutor()),
+  ven: () => import("./venice-web.ts").then((m) => new m.VeniceWebExecutor()), // Alias
+  "notion-web": () => import("./notion-web.ts").then((m) => new m.NotionWebExecutor()),
+  nw: () => import("./notion-web.ts").then((m) => new m.NotionWebExecutor()), // Alias
+  promptql: () => import("./promptql.ts").then((m) => new m.PromptQlExecutor()),
+  pql: () => import("./promptql.ts").then((m) => new m.PromptQlExecutor()), // Alias
+  "v0-vercel-web": () => import("./v0-vercel-web.ts").then((m) => new m.V0VercelWebExecutor()),
+  v0: () => import("./v0-vercel-web.ts").then((m) => new m.V0VercelWebExecutor()), // Alias
+  "kimi-web": () => import("./kimi-web.ts").then((m) => new m.KimiWebExecutor()),
+  "kimi-coding-apikey": () =>
+    import("./kimi.ts").then((m) => new m.KimiExecutor("kimi-coding-apikey")), // Legacy alias
+  "kimi-coding": () => import("./kimi.ts").then((m) => new m.KimiExecutor()), // Alias
+  moonshot: () => import("./moonshot.ts").then((m) => new m.MoonshotExecutor()),
+  kimi: () => import("./moonshot.ts").then((m) => new m.MoonshotExecutor("kimi")), // Hidden legacy Moonshot provider id
+  cheaperinference: () =>
+    import("./cheaperinference.ts").then((m) => new m.CheaperInferenceExecutor()),
+  cinf: () =>
+    import("./cheaperinference.ts").then(
+      (m) => new m.CheaperInferenceExecutor("cheaperinference")
+    ), // Alias
+  "doubao-web": () => import("./doubao-web.ts").then((m) => new m.DoubaoWebExecutor()),
+  db: () => import("./doubao-web.ts").then((m) => new m.DoubaoWebExecutor()), // Alias
+  "qwen-web": () => import("./qwen-web.ts").then((m) => new m.QwenWebExecutor()),
+  raycast: () => import("./raycast.ts").then((m) => new m.RaycastExecutor()),
+  rc: () => import("./raycast.ts").then((m) => new m.RaycastExecutor()), // Alias
+  "hailuo-web": () => import("./hailuo-web.ts").then((m) => new m.HailuoWebExecutor()),
+  "zai-web": () => import("./zai-web.ts").then((m) => new m.ZaiWebExecutor()),
+  zw: () => import("./zai-web.ts").then((m) => new m.ZaiWebExecutor()), // Alias
+  theoldllm: () => import("./theoldllm.ts").then((m) => new m.TheOldLlmExecutor()),
+  tllm: () => import("./theoldllm.ts").then((m) => new m.TheOldLlmExecutor()), // Alias
+  chipotle: () => import("./chipotle.ts").then((m) => new m.ChipotleExecutor()),
+  pepper: () => import("./chipotle.ts").then((m) => new m.ChipotleExecutor()), // Alias
+  lmarena: () => import("./lmarena.ts").then((m) => new m.LMArenaExecutor()),
+  lma: () => import("./lmarena.ts").then((m) => new m.LMArenaExecutor()), // Alias
+  "grok-cli": () => import("./grok-cli.ts").then((m) => new m.GrokCliExecutor()),
+  gc: () => import("./grok-cli.ts").then((m) => new m.GrokCliExecutor()), // Alias
+  "codebuddy-cn": () => import("./codebuddy-cn.ts").then((m) => new m.CodeBuddyCnExecutor()),
+  cbcn: () => import("./codebuddy-cn.ts").then((m) => new m.CodeBuddyCnExecutor()), // Alias for codebuddy-cn
+  "zenmux-free": () => import("./zenmux-free.ts").then((m) => new m.ZenmuxFreeExecutor()),
+  "cloudflare-playground": () =>
+    import("./cloudflare-playground.ts").then((m) => new m.CloudflarePlaygroundExecutor()),
+  cfp: () =>
+    import("./cloudflare-playground.ts").then((m) => new m.CloudflarePlaygroundExecutor()), // Alias for cloudflare-playground
+  "tinycms-web": () => import("./tinycms.ts").then((m) => new m.TinyCmsExecutor()),
+  tcw: () => import("./tinycms.ts").then((m) => new m.TinyCmsExecutor()), // Alias
+  hyperagent: () => import("./hyperagent.ts").then((m) => new m.HyperAgentExecutor()),
+  ha: () => import("./hyperagent.ts").then((m) => new m.HyperAgentExecutor()), // Alias
+  zmf: () => import("./zenmux-free.ts").then((m) => new m.ZenmuxFreeExecutor()), // Alias for zenmux-free
+  auggie: () => import("./auggie.ts").then((m) => new m.AuggieExecutor()),
+  xai: () => import("./xai.ts").then((m) => new m.XaiExecutor()),
+  "xai-oauth": () => import("./xai.ts").then((m) => new m.XaiExecutor("xai-oauth")),
+  xao: () => import("./xai.ts").then((m) => new m.XaiExecutor("xai-oauth")),
+  qw: () => import("./qwen-web.ts").then((m) => new m.QwenWebExecutor()), // Alias
+  "conol-web": () => import("./conol-web.ts").then((m) => new m.ConolWebExecutor()),
+  cnl: () => import("./conol-web.ts").then((m) => new m.ConolWebExecutor()), // Alias
 };
 
-// Bootstrap: register every built-in in the ExecutorRegistry. registerExecutor
-// throws on duplicates, so an alias collision fails at module load, exactly as
-// loudly as a duplicate object key would have failed at lint time.
-for (const [alias, executor] of Object.entries(executors) as [string, BaseExecutor][]) {
-  registerExecutor(alias, executor);
+// Bootstrap: declare every built-in alias in the ExecutorRegistry. Duplicate
+// aliases fail at module load, exactly as loudly as a duplicate object key
+// would have failed at lint time. Instances materialize on first use (#11220).
+for (const [alias, load] of Object.entries(lazyExecutors)) {
+  registerLazyExecutor(alias, load);
 }
 
 const defaultCache = new Map();
@@ -267,9 +239,9 @@ const CHAT_UNSUPPORTED_CLOUD_AGENT_PROVIDERS = new Set(["jules"]);
 // providers must be executed through /v1/search, never the chat-completions path.
 const CHAT_UNSUPPORTED_SEARCH_PROVIDERS = new Set(Object.keys(SEARCH_PROVIDERS));
 
-export function getExecutor(provider) {
-  const registered = getRegisteredExecutor(provider);
-  if (registered) return registered;
+export async function getExecutor(provider: string): Promise<BaseExecutor> {
+  const loaded = await loadRegisteredExecutor(provider);
+  if (loaded) return loaded;
   if (CHAT_UNSUPPORTED_CLOUD_AGENT_PROVIDERS.has(provider)) {
     const err = new Error(
       `Provider "${provider}" is a cloud-agent provider and does not support direct chat completions; use the Cloud Agents task API instead.`
@@ -285,77 +257,19 @@ export function getExecutor(provider) {
     throw err;
   }
   if (!defaultCache.has(provider)) defaultCache.set(provider, new DefaultExecutor(provider));
-  return defaultCache.get(provider);
+  return defaultCache.get(provider)!;
 }
 
-export function hasSpecializedExecutor(provider) {
+export function hasSpecializedExecutor(provider: string): boolean {
   return hasRegisteredExecutor(provider);
 }
 
-export { registerExecutor, listExecutorAliases } from "./registry.ts";
-
+export {
+  registerExecutor,
+  registerLazyExecutor,
+  listExecutorAliases,
+} from "./registry.ts";
+// Value re-export: base.ts is already eager (DefaultExecutor extends it), and
+// scripts/check/check-known-symbols.ts reads this export from the module.
 export { BaseExecutor } from "./base.ts";
-export { AntigravityExecutor } from "./antigravity.ts";
-export { GithubExecutor } from "./github.ts";
-export { QoderExecutor } from "./qoder.ts";
-export { KiroExecutor } from "./kiro.ts";
-export { CodexExecutor } from "./codex.ts";
-export { CursorExecutor } from "./cursor.ts";
-export { TraeExecutor } from "./trae.ts";
 export { DefaultExecutor } from "./default.ts";
-export { BedrockExecutor } from "./bedrock.ts";
-export { GlmExecutor } from "./glm.ts";
-export { PollinationsExecutor } from "./pollinations.ts";
-export { CloudflareAIExecutor } from "./cloudflare-ai.ts";
-export { OpencodeExecutor } from "./opencode.ts";
-export { CliproxyapiExecutor } from "./cliproxyapi.ts";
-export { DarioExecutor } from "./dario.ts";
-export { NineRouterExecutor } from "./ninerouter.ts";
-export { VertexExecutor } from "./vertex.ts";
-export { PerplexityWebExecutor } from "./perplexity-web.ts";
-export { GrokWebExecutor } from "./grok-web.ts";
-export { GeminiWebExecutor } from "./gemini-web.ts";
-export { KieExecutor } from "./kie.ts";
-export { ChatGptWebExecutor } from "./chatgpt-web.ts";
-export { BlackboxWebExecutor } from "./blackbox-web.ts";
-export { MuseSparkWebExecutor } from "./muse-spark-web.ts";
-export { AzureOpenAIExecutor } from "./azure-openai.ts";
-export { AzureAiExecutor } from "./azure-ai.ts";
-export { CommandCodeExecutor } from "./commandCode.ts";
-export { GitlabExecutor } from "./gitlab.ts";
-export { NlpCloudExecutor } from "./nlpcloud.ts";
-export { DevinDesktopExecutor } from "./devin-desktop.ts";
-export { ZedHostedExecutor } from "./zed-hosted.ts";
-export { DevinCliExecutor } from "./devin-cli.ts";
-export { DevinCliAgenticExecutor } from "./devin-cli-agentic.ts";
-export { AuggieExecutor } from "./auggie.ts";
-export { CopilotWebExecutor } from "./copilot-web.ts";
-export { CopilotM365WebExecutor } from "./copilot-m365-web.ts";
-export { MicrosoftDesignerWebExecutor } from "./microsoft-designer-web.ts";
-export { AdobeFireflyExecutor } from "./adobe-firefly.ts";
-export { VeoAIFreeWebExecutor } from "./veoaifree-web.ts";
-export { DuckDuckGoWebExecutor } from "./duckduckgo-web.ts";
-export { FeloWebExecutor } from "./felo-web.ts";
-export { ClaudeWebExecutor } from "./claude-web.ts";
-export { DeepSeekWebExecutor } from "./deepseek-web.ts";
-export { DeepSeekWebWithAutoRefreshExecutor } from "./deepseek-web-with-auto-refresh.ts";
-export { AdaptaWebExecutor } from "./adapta-web.ts";
-export { YuanbaoWebExecutor } from "./yuanbao-web.ts";
-export { T3ChatWebExecutor } from "./t3-chat-web.ts";
-export { InnerAiExecutor } from "./inner-ai.ts";
-export { QwenWebExecutor } from "./qwen-web.ts";
-export { HailuoWebExecutor } from "./hailuo-web.ts";
-export { TheOldLlmExecutor } from "./theoldllm.ts";
-export { ChipotleExecutor } from "./chipotle.ts";
-export { LMArenaExecutor } from "./lmarena.ts";
-export { GrokCliExecutor } from "./grok-cli.ts";
-export { CodeBuddyCnExecutor } from "./codebuddy-cn.ts";
-export { ZenmuxFreeExecutor } from "./zenmux-free.ts";
-export { CloudflarePlaygroundExecutor } from "./cloudflare-playground.ts";
-export { TinyCmsExecutor } from "./tinycms.ts";
-export { HyperAgentExecutor } from "./hyperagent.ts";
-export { XaiExecutor } from "./xai.ts";
-export { MoonshotExecutor } from "./moonshot.ts";
-export { CheaperInferenceExecutor } from "./cheaperinference.ts";
-export { PromptQlExecutor } from "./promptql.ts";
-export { ConolWebExecutor } from "./conol-web.ts";

open-sse/executors/opencode.ts

@@ -895,10 +895,21 @@ export class OpencodeExecutor extends BaseExecutor {
       const mb = modifiedBody as Record<string, unknown>;
       const parsed = parseEffortLevel(model);
       if (parsed) {
-        mb.model = parsed.baseModel;
-        if (mb.reasoning_effort === undefined) {
-          mb.reasoning_effort = parsed.effort;
+        const deepseekFamily =
+          parsed.baseModel === "deepseek-v4-pro" || parsed.baseModel === "deepseek-v4-flash";
+        if (deepseekFamily) {
+          // DeepSeek via opencode-go proxies the native DeepSeek contract, which
+          // accepts a flat reasoning_effort field (#4647).
+          mb.model = parsed.baseModel;
+          if (mb.reasoning_effort === undefined) {
+            mb.reasoning_effort = parsed.effort;
+          }
         }
+        // #10788: every other family's ONLY native effort mechanism is the
+        // -<tier> suffix in the model id itself (the ids `opencode models
+        // opencode-go --verbose` lists). The opencode-go ChatCompletionRequest
+        // carries no flat reasoning_effort field, so rewriting to the base id
+        // silently dropped the tier — forward the aliased id verbatim instead.
       }
     }
     // #1543 / upstream PR #1099: thinking-mode upstreams routed through OpenCode

open-sse/executors/registry.ts

@@ -28,11 +28,46 @@ export function getRegisteredExecutor(alias: string): BaseExecutor | undefined {
   return registry.get(alias);
 }
 
-export function hasRegisteredExecutor(alias: string): boolean {
-  return registry.has(alias);
+// ── #11220: lazy registration ───────────────────────────────────────────────
+// Aliases may register a deferred loader instead of an instance. The alias and
+// its registration ORDER are declared eagerly — hasRegisteredExecutor() and
+// listExecutorAliases() stay synchronous and the golden snapshot keeps its
+// shape — while the class import + construction happen on first use. A
+// completed load caches into `registry`, so later resolution is identical to a
+// static registration.
+const lazyLoaders = new Map<string, () => Promise<BaseExecutor>>();
+const lazyInFlight = new Map<string, Promise<BaseExecutor>>();
+
+export function registerLazyExecutor(alias: string, load: () => Promise<BaseExecutor>): void {
+  if (registry.has(alias) || lazyLoaders.has(alias)) {
+    throw new Error(`executor alias already registered: "${alias}"`);
+  }
+  lazyLoaders.set(alias, load);
 }
 
-/** All registered aliases, in registration order. */
+export function loadRegisteredExecutor(alias: string): Promise<BaseExecutor> | undefined {
+  const cached = registry.get(alias);
+  if (cached) return Promise.resolve(cached);
+  const load = lazyLoaders.get(alias);
+  if (!load) return undefined;
+  let inFlight = lazyInFlight.get(alias);
+  if (!inFlight) {
+    inFlight = load().then((executor) => {
+      registerExecutor(alias, executor);
+      lazyLoaders.delete(alias);
+      lazyInFlight.delete(alias);
+      return executor;
+    });
+    lazyInFlight.set(alias, inFlight);
+  }
+  return inFlight;
+}
+
+export function hasRegisteredExecutor(alias: string): boolean {
+  return registry.has(alias) || lazyLoaders.has(alias);
+}
+
+/** All registered aliases — static and lazy — in registration order. */
 export function listExecutorAliases(): string[] {
-  return [...registry.keys()];
+  return [...registry.keys(), ...lazyLoaders.keys()];
 }

open-sse/handlers/chatCore/cliproxyModelMapping.ts

@@ -14,12 +14,12 @@
 type ExecutorInput = {
   model: string;
   body: unknown;
-  [key: string]: unknown;
 };
 
+// No index signature: executors (BaseExecutor subclasses) must satisfy this
+// structurally, and class instances don't carry index signatures.
 type ExecutorLike = {
   execute: (input: ExecutorInput) => Promise<unknown>;
-  [key: string]: unknown;
 };
 
 export type CliproxyapiModelMapping = Record<string, unknown> | null | undefined;

open-sse/handlers/chatCore/cliproxyapiCredentials.ts

@@ -19,12 +19,12 @@ import type { ProviderCredentials } from "../../executors/base.ts";
 
 type ExecutorInput = {
   credentials: ProviderCredentials;
-  [key: string]: unknown;
 };
 
+// No index signature: executors (BaseExecutor subclasses) must satisfy this
+// structurally, and class instances don't carry index signatures.
 type ExecutorLike = {
   execute: (input: ExecutorInput) => Promise<unknown>;
-  [key: string]: unknown;
 };
 
 /**

open-sse/handlers/chatCore/executorProxy.ts

@@ -76,12 +76,15 @@ async function loadCliproxyapiSettings(): Promise<{
  * dedicated-credential wrappers applied. Used by the direct `cliproxyapi` leg
  * and the CLIProxyAPI branch of `fallback`.
  */
-function resolveCliproxyapiExecutor(
+async function resolveCliproxyapiExecutor(
   cliproxyapiModelMapping: Record<string, unknown> | null,
   dedicatedApiKey: string | null
 ) {
   return wrapExecutorWithCliproxyapiCredentials(
-    wrapExecutorWithCliproxyapiModelMapping(getExecutor("cliproxyapi"), cliproxyapiModelMapping),
+    wrapExecutorWithCliproxyapiModelMapping(
+      await getExecutor("cliproxyapi"),
+      cliproxyapiModelMapping
+    ),
     dedicatedApiKey
   );
 }
@@ -138,7 +141,7 @@ export async function resolveExecutorWithProxy(
   // backend on specific failures. The backend defaults to CLIProxyAPI so every
   // pre-existing fallback config behaves exactly as before; fallbackBackend
   // === "dario" opts the retry leg over to Dario instead.
-  const nativeExec = getExecutor(prov);
+  const nativeExec = await getExecutor(prov);
   const fallbackBackend: FallbackBackend = cfg.fallbackBackend;
   const { fallbackCodes, dedicatedApiKey } = await loadCliproxyapiSettings();
 
@@ -146,8 +149,8 @@ export async function resolveExecutorWithProxy(
   // the native leg must keep seeing the original, unmapped model.
   const proxyExec =
     fallbackBackend === "dario"
-      ? getExecutor("dario")
-      : resolveCliproxyapiExecutor(cfg.cliproxyapiModelMapping, dedicatedApiKey);
+      ? await getExecutor("dario")
+      : await resolveCliproxyapiExecutor(cfg.cliproxyapiModelMapping, dedicatedApiKey);
   const backendLabel = fallbackBackend === "dario" ? "Dario" : "CLIProxyAPI";
   const isRetryableStatus = (s: number) => fallbackCodes.includes(s) || s === 0;
 

open-sse/handlers/rerank.ts

@@ -201,6 +201,7 @@ export async function handleRerank({
   connectionId = null,
   apiKeyId = null,
   apiKeyName = null,
+  resolvedProvider = null,
 }) {
   const startTime = Date.now();
   if (!model) return errorResponse(400, "model is required");
@@ -210,7 +211,8 @@ export async function handleRerank({
   }
 
   const { provider: providerId, model: modelId } = parseRerankModel(model);
-  const providerConfig = providerId ? getRerankProvider(providerId) : null;
+  const providerConfig =
+    resolvedProvider || (providerId ? getRerankProvider(providerId) : null);
 
   if (!providerConfig) {
     const availableProviders = Object.keys(RERANK_PROVIDERS).join(", ");
@@ -219,10 +221,13 @@ export async function handleRerank({
       `No rerank provider found for model "${model}". Available: ${availableProviders}`
     );
   }
+  // When a derived/generic provider is injected, its id is authoritative for
+  // logging and cost attribution even though parseRerankModel returned null.
+  const effectiveProviderId = providerConfig.id || providerId;
 
   const token = credentials?.apiKey || credentials?.accessToken;
   if (!token) {
-    return errorResponse(401, `No credentials for rerank provider: ${providerId}`);
+    return errorResponse(401, `No credentials for rerank provider: ${effectiveProviderId}`);
   }
 
   const requestBody = transformRequestForProvider(providerConfig, {
@@ -275,8 +280,8 @@ export async function handleRerank({
         method: "POST",
         path: "/v1/rerank",
         status: res.status,
-        model: `${providerId}/${modelId}`,
-        provider: providerId,
+        model: `${effectiveProviderId}/${modelId}`,
+        provider: effectiveProviderId,
         connectionId: connectionId || undefined,
         duration: Date.now() - startTime,
         requestBody,
@@ -296,14 +301,14 @@ export async function handleRerank({
     });
 
     const searchUnits = Number(result?.meta?.billed_units?.search_units) || 0;
-    const costUsd = await calculateModalCost("rerank", providerId, modelId, { searchUnits });
+    const costUsd = await calculateModalCost("rerank", effectiveProviderId, modelId, { searchUnits });
 
     saveCallLog({
       method: "POST",
       path: "/v1/rerank",
       status: 200,
-      model: `${providerId}/${modelId}`,
-      provider: providerId,
+      model: `${effectiveProviderId}/${modelId}`,
+      provider: effectiveProviderId,
       connectionId: connectionId || undefined,
       duration: Date.now() - startTime,
       tokens: { prompt_tokens: 0, completion_tokens: 0 },
@@ -315,7 +320,7 @@ export async function handleRerank({
 
     const headers = new Headers({ ...CORS_HEADERS, "Content-Type": "application/json" });
     attachOmniRouteMetaHeaders(headers, {
-      provider: providerId,
+      provider: effectiveProviderId,
       model: modelId,
       costUsd,
       latencyMs: Date.now() - startTime,

open-sse/handlers/videoGeneration.ts

@@ -374,7 +374,7 @@ async function handleVertexVeoGeneration({ model, body, credentials, log }) {
  * Submits an AnimateDiff or SVD workflow, polls for completion, fetches output video
  */
 async function handleVeoAiFreeVideoGeneration({ model, provider, body, credentials, log }) {
-  const executor = getExecutor(provider);
+  const executor = await getExecutor(provider);
   if (!executor) {
     return { success: false, status: 400, error: `Unknown video provider: ${provider}` };
   }

open-sse/services/autoCombo/builtinCatalog.ts

@@ -44,12 +44,6 @@ export const AUTO_TEMPLATE_VARIANTS: Record<string, AutoVariant | undefined> = {
   "auto/claude-opus": "smart",
   "auto/claude-sonnet": "coding",
   "auto/best-free": "cheap",
-  // Subscription-first routing (see `subscriptionLadder.ts`). `auto/subscription`
-  // maps to no weight variant on purpose: its pool is already restricted to
-  // plan-included connections, so the scorer should rank them on merit rather
-  // than bias toward cheap/fast within an allowance the operator already paid for.
-  "auto/subscription": undefined,
-  "auto/thrifty": "cheap",
   // Chaos mode — parallel dispatch to top-N stable models
   "auto/best-chaos": "chaos",
   "auto/chaos": "chaos",
@@ -73,18 +67,6 @@ export const AUTO_SUFFIX_VARIANTS: string[] = [
   "auto/multimodal",
 ];
 
-/**
- * Flat `auto/*` ids that carry a tier overlay even though they are not written
- * in `<category>:<tier>` form. `auto/best-free` established the pattern; the
- * two subscription-first ids reuse it so a caller can ask for the behavior
- * without also having to pick a category.
- */
-export const FLAT_TIER_OVERLAY_IDS: Record<string, AutoTier> = {
-  "auto/best-free": "free",
-  "auto/subscription": "subscription",
-  "auto/thrifty": "thrifty",
-};
-
 type ResolvedAutoVariant =
   { recognized: true; variant: AutoVariant | undefined } | { recognized: false };
 
@@ -217,9 +199,8 @@ export async function createBuiltinAutoCombo(
   }
 
   if ("variant" in spec && spec.variant !== undefined) {
-    const overlayTier = FLAT_TIER_OVERLAY_IDS[modelStr];
     const virtualCombo = await materialize(spec.variant, {
-      ...(overlayTier ? { tier: overlayTier } : {}),
+      ...(modelStr === "auto/best-free" ? { tier: "free" as const } : {}),
     });
     virtualCombo.name = modelStr;
     virtualCombo.id = modelStr;
@@ -230,11 +211,7 @@ export async function createBuiltinAutoCombo(
   // auto/best-chat, auto/pro-chat) still materialize via the default
   // (unconstrained) virtual combo rather than throwing "Unknown built-in".
   if (Object.prototype.hasOwnProperty.call(AUTO_TEMPLATE_VARIANTS, modelStr)) {
-    const overlayTier = FLAT_TIER_OVERLAY_IDS[modelStr];
-    const virtualCombo = await materialize(
-      undefined,
-      overlayTier ? { tier: overlayTier } : undefined
-    );
+    const virtualCombo = await materialize(undefined);
     virtualCombo.name = modelStr;
     virtualCombo.id = modelStr;
     return virtualCombo;

open-sse/services/autoCombo/connectionBilling.ts

@@ -1,100 +0,0 @@
-/**
- * Pure classifier over the curated connection-billing catalog
- * (`open-sse/config/connectionBillingCatalog.ts`).
- *
- * Kept dependency-light on purpose — the same constraint `paidModelFilter.ts`
- * and `strictZeroCostFilter.ts` state in their own headers — so subscription
- * routing is unit-testable without seeding the DB or the virtual factory. No
- * provider name appears in this file: a connection is classified purely from
- * the catalog plus the two facts the caller already has (`provider`,
- * `authType`), so curating a new provider needs no code change here.
- */
-import {
-  CONNECTION_BILLING_CATALOG,
-  type ConnectionBillingClass,
-  type ConnectionBillingEntry,
-  type ConnectionOverageBehavior,
-} from "@omniroute/open-sse/config/connectionBillingCatalog.ts";
-import { SYNTHETIC_NOAUTH_CONNECTION_ID } from "./resilienceCandidateFilter";
-
-/** The minimum a caller must know about a connection to classify it. */
-export interface BillableConnection {
-  provider: string;
-  /** `provider_connections.auth_type` — `oauth` / `apikey` / `cookie` / … */
-  authType?: string | null;
-  /** Connection id; the synthetic no-auth sentinel classifies as `keyless`. */
-  connectionId?: string | null;
-}
-
-export interface ConnectionBillingVerdict {
-  billing: ConnectionBillingClass;
-  overage: ConnectionOverageBehavior;
-  reason: string;
-}
-
-const UNKNOWN_VERDICT: ConnectionBillingVerdict = {
-  billing: "unknown",
-  overage: "unknown",
-  reason: "No curated billing entry for this provider/authType — assumed metered.",
-};
-
-const KEYLESS_VERDICT: ConnectionBillingVerdict = {
-  billing: "keyless",
-  overage: "hard-stop",
-  reason:
-    "Synthetic no-auth connection: no credential exists, so no request against it can be billed.",
-};
-
-/**
- * Classify one connection.
- *
- * Resolution order, first match wins:
- *   1. the synthetic no-auth sentinel → `keyless` (no credential can be billed);
- *   2. a catalog entry matching BOTH provider and `authType`;
- *   3. a provider-wide catalog entry (no `authType` declared);
- *   4. otherwise `unknown`.
- *
- * `unknown` is never treated as free by any caller — `isPlanIncluded()` below
- * returns false for it, so an uncurated provider stays outside the
- * subscription rung until someone curates it deliberately.
- */
-export function classifyConnectionBilling(
-  connection: BillableConnection,
-  catalog: readonly ConnectionBillingEntry[] = CONNECTION_BILLING_CATALOG
-): ConnectionBillingVerdict {
-  if (connection.connectionId === SYNTHETIC_NOAUTH_CONNECTION_ID) return KEYLESS_VERDICT;
-
-  const provider = connection.provider;
-  if (!provider) return UNKNOWN_VERDICT;
-
-  const providerEntries = catalog.filter((entry) => entry.provider === provider);
-  if (providerEntries.length === 0) return UNKNOWN_VERDICT;
-
-  const authType = typeof connection.authType === "string" ? connection.authType : null;
-  const authMatch = authType
-    ? providerEntries.find((entry) => entry.authType === authType)
-    : undefined;
-  const entry = authMatch ?? providerEntries.find((entry) => entry.authType === undefined);
-  if (!entry) return UNKNOWN_VERDICT;
-
-  return { billing: entry.billing, overage: entry.overage, reason: entry.reason };
-}
-
-/**
- * True when serving a request through this connection consumes an allowance
- * the operator already pays for, rather than adding incremental spend.
- * `keyless` qualifies: it costs nothing by construction.
- */
-export function isPlanIncluded(verdict: ConnectionBillingVerdict): boolean {
-  return verdict.billing === "subscription" || verdict.billing === "keyless";
-}
-
-/**
- * True when exhausting this connection's allowance cannot start costing money.
- * The strict `auto/subscription` grouping admits nothing else: an operator who
- * asked never to spend extra must not be surprised by a provider that meters
- * past the plan, nor by one whose terms simply are not established.
- */
-export function isOverageSafe(verdict: ConnectionBillingVerdict): boolean {
-  return verdict.overage === "hard-stop";
-}

open-sse/services/autoCombo/freeAccessQuota.ts

@@ -22,7 +22,6 @@ import {
 import { getCachedProviderConnections } from "@/lib/db/readCache";
 import { defaultLogger as log } from "@omniroute/open-sse/utils/logger";
 import type { FreeAccessState } from "./strictZeroCostFilter";
-import { isStateStaleForReset } from "./subscriptionLadder";
 
 const USAGE_FETCHER_PROVIDER_SET = new Set<string>(USAGE_FETCHER_PROVIDERS);
 
@@ -192,13 +191,7 @@ export function resolveFreeAccessState(
 
   const key = cacheKey(provider, connectionId);
   const entry = cache.get(key);
-  // Subscription-first routing (decision 3): an entry whose own `resetAt` has
-  // already passed describes a quota window that no longer exists, so it is
-  // stale REGARDLESS of its age. Without this, a plan that refilled at
-  // midnight keeps reading EXHAUSTED until the TTL happens to lapse, and
-  // routing stays on paid rungs for no reason. See `subscriptionLadder.ts`.
-  const resetElapsed = entry !== undefined && isStateStaleForReset(entry.state);
-  const fresh = entry && !resetElapsed && Date.now() - entry.fetchedAtMs <= ttlMs();
+  const fresh = entry && Date.now() - entry.fetchedAtMs <= ttlMs();
   if (!fresh) {
     void refresh(provider, connectionId);
   }

open-sse/services/autoCombo/subscriptionLadder.ts

@@ -1,411 +0,0 @@
-/**
- * Subscription-first routing: the rung model, its two groupings, and the
- * reset re-entry rules.
- *
- * OmniRoute already answers "is this model free?" (`paidModelFilter.ts`) and
- * "can this connection ever bill me?" (`strictZeroCostFilter.ts`). Both fail
- * CLOSED — an exhausted free pool is an empty pool, never a step up to a paid
- * option. And every paid-side mechanism (`cost-optimized`, `budgetCap`,
- * the `cost-saver` mode pack) is tier-agnostic. Neither side answers:
- *
- *   "use the quota I already pay for; when it runs out either stop, or step up
- *    one rung at a time; and come back the moment it resets."
- *
- * This module supplies both halves of that, sharing one rung model:
- *
- *   - `filterSubscriptionOnlyCandidates` — the strict grouping (`auto/subscription`).
- *     Rung 0 only, overage-safe connections only, verified live. Fails CLOSED.
- *   - `orderPoolByRung` — the escalating grouping (`auto/thrifty`). All rungs,
- *     ordered, with exhausted rungs gated out. Fails OPEN, one rung at a time.
- *
- * Design mirrors `strictZeroCostFilter.ts` deliberately: pure functions, the
- * live quota lookup injected as a synchronous resolver, no DB or network
- * import, and the SAME connection-safety invariant — every connection in a
- * candidate's `allowedConnectionIds` is verified INDIVIDUALLY and the array is
- * rewritten to exactly the surviving subset, never the full original list.
- * `autoStrategy.ts` enforces `allowedConnectionIds` as a hard allowlist before
- * selecting a connection at dispatch, so rewriting it here is sufficient to
- * make "verified" and "actually used" the same set by construction.
- */
-import {
-  classifyConnectionBilling,
-  isOverageSafe,
-  type BillableConnection,
-} from "./connectionBilling";
-import type { ConnectionBillingEntry } from "@omniroute/open-sse/config/connectionBillingCatalog.ts";
-import type { FreeAccessState } from "./strictZeroCostFilter";
-
-/**
- * Rungs in escalation order. Index is the ordering key; membership is decided
- * by `assignRung` below.
- *
- * The rungs differ in more than price — each has its OWN exhaustion signal,
- * which is why this is not just a sort:
- *
- *   subscription / keyless / free → exhausted on QUOTA (observable, tracked)
- *   cheap / premium              → exhausted on BUDGET (no quota exists; a paid
- *                                  connection serves forever)
- */
-export const RUNG_ORDER = ["subscription", "keyless", "free", "cheap", "premium"] as const;
-
-export type LadderRung = (typeof RUNG_ORDER)[number];
-
-/** Rungs whose exhaustion is observable from provider quota state. */
-const QUOTA_BEARING_RUNGS: ReadonlySet<LadderRung> = new Set<LadderRung>([
-  "subscription",
-  "keyless",
-  "free",
-]);
-
-export function rungIndex(rung: LadderRung): number {
-  return RUNG_ORDER.indexOf(rung);
-}
-
-/** A candidate as this module needs to see it — a structural subset of
- * `VirtualAutoComboCandidate` (`virtualFactory.ts`), so this file has no
- * dependency on that module's full type. */
-export interface LadderCandidate {
-  provider: string;
-  model: string;
-  connectionId: string | null;
-  allowedConnectionIds?: string[];
-}
-
-export interface LadderOptions {
-  /** Master switch. When false every exported filter is the identity function
-   * — the same off-by-default contract `filterPaidOnlyCandidates` holds. */
-  enabled: boolean;
-  /**
-   * Live allowance/quota state for ONE (provider, connection) pair, resolved
-   * from the cache in `freeAccessQuota.ts`. Synchronous by design: nothing in
-   * a candidate-pool build may await a network call.
-   *
-   * `undefined` means "no usage adapter for this provider, or nothing fresh
-   * cached". The two groupings interpret that OPPOSITELY on purpose — see
-   * `admitUnknownQuota` below.
-   */
-  resolveFreeAccessState: (provider: string, connectionId: string) => FreeAccessState | undefined;
-  /**
-   * `authType` for a connection id (`provider_connections.auth_type`), needed
-   * to classify billing. Unknown ids resolve to `null` → the provider-wide
-   * catalog entry, or `unknown` billing.
-   */
-  resolveAuthType: (connectionId: string) => string | null;
-  /** Economic tier of a (provider, model) pair — `classifyTier()` in
-   * production, injected so this module needs no registry/pricing import. */
-  resolveEconomicTier: (provider: string, model: string) => "free" | "cheap" | "premium";
-  /**
-   * Remaining-percent at or below which a quota-bearing connection counts as
-   * exhausted. Default 2, matching `quotaPreflight.defaultThresholdPercent`
-   * (`src/lib/resilience/settings/types.ts`) so the two agree.
-   */
-  exitCutoffPercent?: number;
-  /**
-   * Remaining-percent a quota-bearing connection must EXCEED to be re-admitted
-   * after having been exhausted. Strictly greater than `exitCutoffPercent`;
-   * the gap is the hysteresis band that stops a connection hovering at the
-   * cutoff from oscillating between rungs on consecutive requests. Default 5.
-   */
-  reentryMinRemainingPercent?: number;
-  /** Max age of a `FreeAccessState.checkedAt` before it is treated as stale. */
-  maxStateAgeMs: number;
-  /**
-   * Whether a connection with no usable quota reading is admitted.
-   *
-   *   - `auto/thrifty` passes TRUE: trying a plan-included connection costs
-   *     nothing, and if it turns out to be exhausted the dispatcher's
-   *     fall-through reaches the next rung anyway. Refusing to try it would
-   *     send a request to a PAID rung on missing telemetry — the exact
-   *     outcome the grouping exists to avoid.
-   *   - `auto/subscription` passes FALSE: its promise is that no request can
-   *     cost extra, and an unverifiable connection cannot support that promise.
-   */
-  admitUnknownQuota: boolean;
-  /**
-   * Budget consumed so far on a paid rung, in USD, for the operator's current
-   * budget window. `null`/`undefined` means no spend accounting is available,
-   * in which case paid rungs are NOT budget-gated (they still order after
-   * every plan-included rung). See the spec's open question on the ledger.
-   */
-  resolveRungSpendUsd?: (rung: LadderRung) => number | null;
-  /** Per-rung budget in USD. A rung mapped to 0 is disabled outright. */
-  rungBudgetUsd?: Partial<Record<LadderRung, number>>;
-  /** `now` injection for deterministic tests. */
-  now?: () => number;
-  /** Catalog override for tests; production callers never pass this. */
-  catalog?: readonly ConnectionBillingEntry[];
-}
-
-const DEFAULT_EXIT_CUTOFF_PERCENT = 2;
-const DEFAULT_REENTRY_MIN_REMAINING_PERCENT = 5;
-
-/**
- * Which rung a specific (candidate, connection) pair belongs to.
- *
- * Billing class decides first because it is the fact that actually determines
- * whether money moves; only a genuinely metered connection falls through to
- * the model's economic tier. `unknown` billing is metered by definition
- * (`connectionBilling.ts`), so an uncurated provider lands on a paid rung
- * rather than silently joining the subscription rung.
- */
-export function assignRung(
-  candidate: Pick<LadderCandidate, "provider" | "model">,
-  connection: BillableConnection,
-  options: Pick<LadderOptions, "resolveEconomicTier" | "catalog">
-): LadderRung {
-  const verdict = classifyConnectionBilling(connection, options.catalog);
-  if (verdict.billing === "subscription") return "subscription";
-  if (verdict.billing === "keyless") return "keyless";
-  return options.resolveEconomicTier(candidate.provider, candidate.model);
-}
-
-/**
- * Is this connection's plan allowance usable right now?
- *
- * `hasBeenExhausted` selects which side of the hysteresis band applies: a
- * connection that is currently in play only has to stay above the exit cutoff,
- * while one that already dropped out has to climb back above the (higher)
- * re-entry threshold before it is admitted again.
- */
-export function isQuotaUsable(
-  state: FreeAccessState | undefined,
-  options: Pick<
-    LadderOptions,
-    | "exitCutoffPercent"
-    | "reentryMinRemainingPercent"
-    | "maxStateAgeMs"
-    | "admitUnknownQuota"
-    | "now"
-  >,
-  hasBeenExhausted = false
-): boolean {
-  if (!state) return options.admitUnknownQuota;
-  if (state.status === "EXHAUSTED") return false;
-  if (state.status === "UNKNOWN") return options.admitUnknownQuota;
-
-  const now = (options.now ?? Date.now)();
-  const checkedAtMs = Date.parse(state.checkedAt);
-  if (!Number.isFinite(checkedAtMs) || now - checkedAtMs > options.maxStateAgeMs) {
-    return options.admitUnknownQuota;
-  }
-
-  if (state.remainingFreeAllowance === null) return options.admitUnknownQuota;
-
-  const exitCutoff = options.exitCutoffPercent ?? DEFAULT_EXIT_CUTOFF_PERCENT;
-  const reentryFloor = Math.max(
-    options.reentryMinRemainingPercent ?? DEFAULT_REENTRY_MIN_REMAINING_PERCENT,
-    exitCutoff
-  );
-  const threshold = hasBeenExhausted ? reentryFloor : exitCutoff;
-  return state.remainingFreeAllowance > threshold;
-}
-
-/**
- * Decision 3 — re-entry after a plan quota resets.
- *
- * A cached state whose own `resetAt` has already passed describes a window
- * that no longer exists. Waiting out the cache TTL before re-reading it is
- * pure lag on the single transition subscription-first routing cares most
- * about, so such an entry is stale REGARDLESS of its age.
- *
- * Consumed by `freeAccessQuota.ts`, which owns the cache; kept here so the
- * rule sits with the rest of the ladder's semantics and is testable without
- * touching the cache.
- */
-export function isStateStaleForReset(
-  state: Pick<FreeAccessState, "resetAt"> | undefined,
-  now: number = Date.now()
-): boolean {
-  if (!state?.resetAt) return false;
-  const resetAtMs = Date.parse(state.resetAt);
-  if (!Number.isFinite(resetAtMs)) return false;
-  return resetAtMs <= now;
-}
-
-/**
- * Decision 3, second half — never hold a plan-included connection in cooldown
- * past the moment its own upstream says the quota is back.
- *
- * The exhausting 429 sets `rateLimitedUntil` from exponential backoff
- * (`baseCooldownMs * 2 ** failureIndex`, `src/sse/services/auth.ts`), which for
- * a subscription connection routinely overshoots the real reset — leaving
- * routing stuck on paid rungs long after the plan refilled.
- *
- * This only ever NARROWS a cooldown, and only when the upstream itself
- * supplied the reset instant. An absent, unparseable, or already-past
- * `resetAt` returns the original cooldown untouched.
- */
-export function clampCooldownToReset(
-  cooldownMs: number,
-  resetAt: string | null | undefined,
-  now: number = Date.now()
-): number {
-  if (!resetAt) return cooldownMs;
-  const resetAtMs = Date.parse(resetAt);
-  if (!Number.isFinite(resetAtMs)) return cooldownMs;
-  const untilResetMs = resetAtMs - now;
-  if (untilResetMs <= 0) return cooldownMs;
-  return Math.min(cooldownMs, untilResetMs);
-}
-
-/** True when a paid rung has consumed its configured budget for the window. */
-export function isRungBudgetExhausted(rung: LadderRung, options: LadderOptions): boolean {
-  const budget = options.rungBudgetUsd?.[rung];
-  if (budget === undefined) return false;
-  if (budget <= 0) return true; // explicitly disabled
-  const spent = options.resolveRungSpendUsd?.(rung);
-  if (spent === null || spent === undefined) return false; // no accounting → not gated
-  return spent >= budget;
-}
-
-/**
- * Connections on a candidate that are usable right now, paired with the rung
- * each one sits on. Quota-bearing rungs are verified per connection; paid
- * rungs have nothing per-connection to verify (they are gated per rung).
- */
-function evaluateConnections(
-  candidate: LadderCandidate,
-  options: LadderOptions,
-  accept: (rung: LadderRung) => boolean
-): { rung: LadderRung; connectionIds: string[] } | null {
-  const connectionIds = candidate.connectionId
-    ? [candidate.connectionId]
-    : (candidate.allowedConnectionIds ?? []);
-  if (connectionIds.length === 0) return null;
-
-  let bestRung: LadderRung | null = null;
-  const usable: string[] = [];
-
-  for (const connectionId of connectionIds) {
-    const rung = assignRung(
-      candidate,
-      {
-        provider: candidate.provider,
-        authType: options.resolveAuthType(connectionId),
-        connectionId,
-      },
-      options
-    );
-    if (!accept(rung)) continue;
-
-    if (QUOTA_BEARING_RUNGS.has(rung)) {
-      const state = options.resolveFreeAccessState(candidate.provider, connectionId);
-      if (!isQuotaUsable(state, options)) continue;
-    } else if (isRungBudgetExhausted(rung, options)) {
-      continue;
-    }
-
-    usable.push(connectionId);
-    // A candidate reachable through several accounts is represented by its
-    // CHEAPEST usable rung: that is the rung a request through it would
-    // actually land on once dispatch picks from the surviving allowlist.
-    if (bestRung === null || rungIndex(rung) < rungIndex(bestRung)) bestRung = rung;
-  }
-
-  if (usable.length === 0 || bestRung === null) return null;
-  return { rung: bestRung, connectionIds: usable };
-}
-
-/** Rewrite a candidate's connection allowlist to the verified subset, keeping
- * the identity-when-nothing-changed contract the sibling filters hold. */
-function withVerifiedConnections<T extends LadderCandidate>(
-  candidate: T,
-  connectionIds: string[]
-): { candidate: T; changed: boolean } {
-  if (candidate.connectionId !== null) return { candidate, changed: false };
-  const original = candidate.allowedConnectionIds ?? [];
-  const isSameSet =
-    original.length === connectionIds.length && connectionIds.every((id) => original.includes(id));
-  if (isSameSet) return { candidate, changed: false };
-  return { candidate: { ...candidate, allowedConnectionIds: connectionIds }, changed: true };
-}
-
-/**
- * `auto/subscription` — the strict grouping. Keeps only candidates servable by
- * a plan-included connection whose overage is a documented hard stop, with
- * live quota headroom verified per connection.
- *
- * Fails CLOSED in every ambiguous case: uncurated provider, unverifiable
- * quota, or an overage that meters to paid. An empty result is the correct,
- * intended answer for an operator who asked never to spend extra — the
- * caller's existing empty-pool path handles it, exactly as `hidePaidModels`
- * already does.
- */
-export function filterSubscriptionOnlyCandidates<T extends LadderCandidate>(
-  pool: T[],
-  options: LadderOptions
-): T[] {
-  if (!options.enabled) return pool;
-
-  const strictOptions: LadderOptions = { ...options, admitUnknownQuota: false };
-  const kept: T[] = [];
-  let changed = false;
-
-  for (const candidate of pool) {
-    const connectionIds = candidate.connectionId
-      ? [candidate.connectionId]
-      : (candidate.allowedConnectionIds ?? []);
-
-    const safe = connectionIds.filter((connectionId) => {
-      const connection: BillableConnection = {
-        provider: candidate.provider,
-        authType: options.resolveAuthType(connectionId),
-        connectionId,
-      };
-      const verdict = classifyConnectionBilling(connection, options.catalog);
-      // `keyless` is plan-included in the ladder's sense but is NOT a
-      // subscription: this grouping is "the plan I pay for", so a no-auth
-      // backend does not belong in it.
-      if (verdict.billing !== "subscription") return false;
-      if (!isOverageSafe(verdict)) return false;
-      const state = strictOptions.resolveFreeAccessState(candidate.provider, connectionId);
-      return isQuotaUsable(state, strictOptions);
-    });
-
-    if (safe.length === 0) {
-      changed = true;
-      continue;
-    }
-    const result = withVerifiedConnections(candidate, safe);
-    if (result.changed) changed = true;
-    kept.push(result.candidate);
-  }
-
-  return changed ? kept : pool;
-}
-
-/**
- * `auto/thrifty` — the escalating grouping. Returns the pool ordered by rung,
- * with candidates whose every connection is exhausted (quota) or whose rung is
- * budget-exhausted removed.
- *
- * Ordering only — the `auto` engine still scores WITHIN the surviving pool, so
- * this decides which rungs are in play, not which candidate wins on one. The
- * combo dispatcher already walks targets in order and falls through on
- * failure, so a runtime exhaustion the preflight did not catch still escalates
- * to the next rung inside the same request.
- *
- * Rung eligibility is recomputed from live state on every pool build and
- * nothing is persisted: there is deliberately no sticky "currently on rung 3"
- * record that could outlive a quota reset and wedge routing on paid rungs.
- */
-export function orderPoolByRung<T extends LadderCandidate>(pool: T[], options: LadderOptions): T[] {
-  if (!options.enabled) return pool;
-
-  const ranked: Array<{ candidate: T; rung: LadderRung; order: number }> = [];
-  for (const [order, candidate] of pool.entries()) {
-    const evaluated = evaluateConnections(candidate, options, () => true);
-    if (!evaluated) continue;
-    const result = withVerifiedConnections(candidate, evaluated.connectionIds);
-    ranked.push({ candidate: result.candidate, rung: evaluated.rung, order });
-  }
-
-  ranked.sort((a, b) => {
-    const byRung = rungIndex(a.rung) - rungIndex(b.rung);
-    // Stable within a rung: preserve the pool's incoming order so the auto
-    // scorer's own ranking is not reshuffled by this overlay.
-    return byRung !== 0 ? byRung : a.order - b.order;
-  });
-
-  return ranked.map((entry) => entry.candidate);
-}

open-sse/services/autoCombo/suffixComposition.ts

@@ -23,20 +23,7 @@ import { isVisionModelId } from "@/shared/constants/visionModels";
 import { isVisionBridgeForcedModel } from "@/shared/constants/visionBridgeDefaults";
 
 export type AutoCategory = "coding" | "reasoning" | "vision" | "chat" | "multimodal";
-export type AutoTier =
-  | "fast"
-  | "cheap"
-  | "floor"
-  | "free"
-  | "reliable"
-  | "pro"
-  // Subscription-first routing. Unlike every tier above, these two narrow by
-  // the CONNECTION's billing class, not the model's price — so they are
-  // applied in `virtualFactory.ts` against live connection state rather than
-  // by `buildAutoCandidateFilter` below, which only sees (provider, model).
-  // See `subscriptionLadder.ts` and `docs/routing/SUBSCRIPTION_LADDER.md`.
-  | "subscription"
-  | "thrifty";
+export type AutoTier = "fast" | "cheap" | "floor" | "free" | "reliable" | "pro";
 
 export const AUTO_CATEGORIES: readonly AutoCategory[] = [
   "coding",
@@ -52,8 +39,6 @@ export const AUTO_TIERS: readonly AutoTier[] = [
   "free",
   "reliable",
   "pro",
-  "subscription",
-  "thrifty",
 ];
 
 const CATEGORY_SET = new Set<string>(AUTO_CATEGORIES);
@@ -99,9 +84,6 @@ export function tierToWeightVariant(tier?: AutoTier): AutoVariant | "reliability
       return "fast";
     case "cheap":
     case "floor":
-    // The ladder already orders plan-included rungs first; within a rung it
-    // should still lean cheap rather than reach for the most expensive model.
-    case "thrifty":
       return "cheap";
     case "reliable":
       return "reliability";
@@ -136,7 +118,8 @@ export function buildAutoCandidateFilter(
       }
       try {
         const caps = getResolvedModelCapabilities({ provider: c.provider, model: c.model });
-        const capable = caps.supportsVision === true || isVisionModelId(c.model);
+        const capable =
+          caps.supportsVision === true || isVisionModelId(c.model);
         if (!capable) return false;
         // #vison-pool: registry entries whose catalog OVERSTATES vision support
         // (opencode-go/opencode-zen/tokenrouter — the backend models are text-only)

open-sse/services/autoCombo/virtualFactory.ts

@@ -28,11 +28,6 @@ import { buildFamilyCandidateFilter, type ModelFamily } from "./modelFamily";
 import { getHiddenModelsByProvider } from "@/models";
 import { getSyncedAvailableModelsByConnection, getCustomModels } from "@/lib/db/models";
 import { filterPaidOnlyCandidates } from "./paidModelFilter";
-import {
-  filterSubscriptionOnlyCandidates,
-  orderPoolByRung,
-  type LadderOptions,
-} from "./subscriptionLadder";
 import { filterStrictZeroCostCandidates, filterTosAvoidCandidates } from "./strictZeroCostFilter";
 import { resolveFreeAccessState } from "./freeAccessQuota";
 import { isModelExcludedByConnection } from "@/domain/connectionModelRules";
@@ -150,98 +145,6 @@ type VirtualAutoCombo = AutoComboConfig & {
 export interface PreparedVirtualAutoComboInputs {
   readonly regularCandidates: readonly VirtualAutoComboCandidate[];
   readonly familyCandidates: readonly VirtualAutoComboCandidate[];
-  /**
-   * `provider_connections.auth_type` per connection id. Subscription-first
-   * routing classifies billing per CONNECTION (`connectionBilling.ts`), and
-   * the candidate pool only carries connection ids — so the auth types are
-   * captured here, during the one bulk connection read this function already
-   * does, instead of re-reading the DB per pool narrowing.
-   */
-  readonly authTypeByConnectionId?: ReadonlyMap<string, string | null>;
-  /** Operator settings for the subscription ladder; absent = feature off. */
-  readonly subscriptionLadder?: SubscriptionLadderSettings;
-}
-
-/**
- * Operator-facing knobs for subscription-first routing. Mirrors the Zod shape
- * in `src/shared/validation/settingsSchemas.ts`.
- *
- * Deliberately TUNING ONLY — there is no `enabled` flag here. `auto/subscription`
- * and `auto/thrifty` are new ids that nothing routes through unless a caller
- * asks for them by name, so requesting the id IS the opt-in. A settings toggle
- * that could switch them off would be actively dangerous: it would leave an id
- * whose whole promise is "plan-included only" quietly serving the full pool,
- * paid models included.
- */
-export interface SubscriptionLadderSettings {
-  exitCutoffPercent?: number;
-  reentryMinRemainingPercent?: number;
-  rungBudgetUsd?: Record<string, number>;
-  /** Staleness bound for a cached quota reading, derived from the existing
-   * `autoRefreshProviderQuotaInterval` exactly as STRICT_ZERO_COST does. */
-  maxStateAgeMs: number;
-}
-
-function readSubscriptionLadderSettings(
-  settings: Record<string, unknown>
-): SubscriptionLadderSettings {
-  const maxStateAgeMs = (Number(settings.autoRefreshProviderQuotaInterval) || 180) * 1000;
-  const raw = settings.subscriptionLadder;
-  if (!raw || typeof raw !== "object") return { maxStateAgeMs };
-  const value = raw as Record<string, unknown>;
-  const numeric = (key: string): number | undefined =>
-    typeof value[key] === "number" && Number.isFinite(value[key] as number)
-      ? (value[key] as number)
-      : undefined;
-  const exitCutoffPercent = numeric("exitCutoffPercent");
-  const reentryMinRemainingPercent = numeric("reentryMinRemainingPercent");
-  return {
-    maxStateAgeMs,
-    ...(exitCutoffPercent === undefined ? {} : { exitCutoffPercent }),
-    ...(reentryMinRemainingPercent === undefined ? {} : { reentryMinRemainingPercent }),
-    ...(value.rungBudgetUsd && typeof value.rungBudgetUsd === "object"
-      ? { rungBudgetUsd: value.rungBudgetUsd as Record<string, number> }
-      : {}),
-  };
-}
-
-/**
- * Build the injected dependencies the pure ladder module needs. Everything it
- * touches is resolved here — the live quota cache, connection auth types, and
- * the economic tier resolver — so `subscriptionLadder.ts` itself stays free of
- * DB, network, and registry imports.
- */
-function buildLadderOptions(
-  prepared: PreparedVirtualAutoComboInputs,
-  tier: "subscription" | "thrifty"
-): LadderOptions {
-  const tuning = prepared.subscriptionLadder;
-  const authTypes = prepared.authTypeByConnectionId;
-  return {
-    enabled: true,
-    resolveFreeAccessState,
-    resolveAuthType: (connectionId: string) => authTypes?.get(connectionId) ?? null,
-    resolveEconomicTier: (provider: string, model: string) => {
-      try {
-        const resolved = classifyTier(provider, model).tier;
-        return resolved === "free" || resolved === "premium" ? resolved : "cheap";
-      } catch {
-        // Same conservative default `safeClassifyTier` uses in suffixComposition.
-        return "cheap";
-      }
-    },
-    maxStateAgeMs: tuning?.maxStateAgeMs ?? 180_000,
-    // The two groupings read a missing quota reading OPPOSITELY on purpose —
-    // see `LadderOptions.admitUnknownQuota`.
-    admitUnknownQuota: tier === "thrifty",
-    ...(tuning?.exitCutoffPercent === undefined
-      ? {}
-      : { exitCutoffPercent: tuning.exitCutoffPercent }),
-    ...(tuning?.reentryMinRemainingPercent === undefined
-      ? {}
-      : { reentryMinRemainingPercent: tuning.reentryMinRemainingPercent }),
-    ...(tuning?.rungBudgetUsd ? { rungBudgetUsd: tuning.rungBudgetUsd } : {}),
-  };
 }
 
 function toExpiryMs(value: unknown): number | null {
@@ -753,15 +656,8 @@ export async function prepareVirtualAutoComboInputs(
   const regularCandidates = buildPreparedPool(false);
   // #6453/#8183: family selectors bypass the reliability-curated no-auth allowlist.
   const familyCandidates = buildPreparedPool(true);
-  // Subscription-first routing inputs, captured from the connection read above
-  // so no later stage has to touch the DB again.
-  const authTypeByConnectionId = new Map<string, string | null>();
-  for (const conn of connections) {
-    authTypeByConnectionId.set(conn.id, typeof conn.authType === "string" ? conn.authType : null);
-  }
-  const subscriptionLadder = readSubscriptionLadderSettings(settings);
   if (!options.includeResolvedCapabilities) {
-    return { regularCandidates, familyCandidates, authTypeByConnectionId, subscriptionLadder };
+    return { regularCandidates, familyCandidates };
   }
 
   // One uninterrupted bulk read of all three capability tables for this prepare only.
@@ -775,8 +671,6 @@ export async function prepareVirtualAutoComboInputs(
   return {
     regularCandidates: await attachPreparedCapabilityValues(regularCandidates, capabilityState),
     familyCandidates: await attachPreparedCapabilityValues(familyCandidates, capabilityState),
-    authTypeByConnectionId,
-    subscriptionLadder,
   };
 }
 
@@ -948,32 +842,6 @@ export async function createVirtualAutoComboFromPrepared(
     }
   }
 
-  // Subscription-first routing (`auto/subscription`, `auto/thrifty`). Applied
-  // AFTER the category/tier narrowing above because, unlike every other tier,
-  // these two select on the connection's billing class and its live quota
-  // state rather than on the model's catalog price — see
-  // `subscriptionLadder.ts` and `docs/routing/SUBSCRIPTION_LADDER.md`.
-  if (spec?.tier === "subscription" || spec?.tier === "thrifty") {
-    const ladderOptions = buildLadderOptions(prepared, spec.tier);
-    const beforeCount = effectivePool.length;
-    effectivePool =
-      spec.tier === "subscription"
-        ? filterSubscriptionOnlyCandidates(effectivePool, ladderOptions)
-        : orderPoolByRung(effectivePool, ladderOptions);
-    if (spec.tier === "subscription" && effectivePool.length === 0 && beforeCount > 0) {
-      // Intended, not a defect: the operator asked for plan-included capacity
-      // only, and right now there is none with verified headroom. Failing
-      // closed here is the entire promise of the id — the caller's existing
-      // empty-pool path turns it into a clear error rather than a silent,
-      // billable fallback.
-      warnEmptyAutoPoolOnce(
-        "auto/subscription",
-        "auto/subscription: no plan-included connection has verified quota headroom; " +
-          "returning an empty pool rather than falling back to paid capacity."
-      );
-    }
-  }
-
   let weights: ScoringWeights = { ...DEFAULT_WEIGHTS };
   let explorationRate = 0.05; // Default exploration rate
   let routerStrategy = "lkgp"; // All auto variants use LKGP

open-sse/services/combo.ts

@@ -926,9 +926,6 @@ async function handleComboChatInner({
   if (activeNativeTurnPin) {
     orderedTargets = applyNativeCodexTurnPin(orderedTargets, activeNativeTurnPin);
     if (orderedTargets.length === 0) {
-      // #11371: quota-share ordering already reserved a winner slot; release it on
-      // this early exit (idempotent).
-      targetResolution.quotaShareRelease?.();
       return errorResponse(
         409,
         "The pinned native Codex turn target is no longer available; the turn cannot be moved to another provider"
@@ -956,8 +953,6 @@ async function handleComboChatInner({
     // no-target failures (silent-stop fix). Threshold of 3 prevents a one-off account
     // wipe from destroying the prompt-cache pin benefit on the next request.
     recordComboFailure(effectiveSessionId, combo.name);
-    // #11371: same early-exit release as the pinned-turn path above.
-    targetResolution.quotaShareRelease?.();
     return errorResponseWithComboDiagnostics(
       404,
       "Combo has no executable targets",
@@ -2846,9 +2841,6 @@ async function handleComboChatInner({
     return await dispatchWithCooldownRetry();
   } finally {
     quotaShareConcurrencyRelease?.();
-    // #11371: release the in-flight slot quota-share ordering reserved for its
-    // winner — the counter must not leak monotonically upward across requests.
-    targetResolution.quotaShareRelease?.();
     // G2: Clean up candidate registry to prevent unbounded memory growth.
     _unregisterExecutionCandidates(_registeredExecutionKeys);
   }

open-sse/services/combo/applyStrategyOrdering.ts

@@ -20,21 +20,6 @@ import {
 } from "./targetSorters.ts";
 import type { ComboLike, ComboLogger, ResolvedComboTarget } from "./types.ts";
 
-/**
- * Result of {@link applyStrategyOrdering}.
- *
- * `quotaShareRelease` carries the idempotent release for the in-flight slot that
- * quota-share ordering reserves for its winner (#11371). It is non-null only when
- * the `quota-share` strategy ran; every other strategy leaves it null. The caller
- * MUST invoke it exactly once when the request settles — selection reserves the
- * slot, so dropping the callback leaks the counter monotonically upward and
- * degenerates P2C into "fewest lifetime dispatches".
- */
-export interface ApplyStrategyOrderingResult {
-  orderedTargets: ResolvedComboTarget[];
-  quotaShareRelease: (() => void) | null;
-}
-
 export interface ApplyStrategyOrderingDeps {
   combo: ComboLike;
   config: Record<string, unknown>;
@@ -60,10 +45,9 @@ export async function applyStrategyOrdering(
   strategy: string,
   initialOrderedTargets: ResolvedComboTarget[],
   deps: ApplyStrategyOrderingDeps
-): Promise<ApplyStrategyOrderingResult> {
+): Promise<ResolvedComboTarget[]> {
   const { combo, config, body, log, apiKeyAllowedConnections, sessionKey } = deps;
   let orderedTargets = initialOrderedTargets;
-  let quotaShareRelease: (() => void) | null = null;
 
   if (strategy === "lkgp") {
     try {
@@ -245,18 +229,14 @@ export async function applyStrategyOrdering(
     const qsModel =
       typeof body?.model === "string" ? body.model : (orderedTargets[0]?.modelStr ?? "");
     const qsMaxConcurrent = await resolveMaxConcurrentByConnection(orderedTargets);
-    const qsSelection = selectQuotaShareTarget(orderedTargets, combo.name, qsModel, Date.now(), {
+    orderedTargets = selectQuotaShareTarget(orderedTargets, combo.name, qsModel, Date.now(), {
       maxConcurrentByConnection: qsMaxConcurrent,
-    });
-    orderedTargets = qsSelection.orderedTargets;
-    // #11371: the reservation made inside selectQuotaShareTarget must outlive this
-    // call — hand the release to the host so it can fire it when the request settles.
-    quotaShareRelease = qsSelection.decrementInflight;
+    }).orderedTargets;
     log.info(
       "COMBO",
       `Quota-share ordering: ${orderedTargets[0]?.modelStr}${orderedTargets[0]?.connectionId ? ` (${orderedTargets[0].connectionId})` : ""} selected (DRR+P2C)`
     );
   }
 
-  return { orderedTargets, quotaShareRelease };
+  return orderedTargets;
 }

open-sse/services/combo/quotaScoring.ts

@@ -261,7 +261,7 @@ function getWindowsMapQuotaWindow(
   );
 }
 
-function resolveQuotaWindowByName(
+export function resolveQuotaWindowByName(
   quota: unknown,
   windowName: ResetWindowName
 ): QuotaWindowSnapshot | null {
@@ -321,8 +321,8 @@ export function scoreResetAwareQuota(
   if (quota.limitReached === true) return { score: -Infinity };
 
   const overallPercentUsed = clamp01(finiteNumberOrNull(quota.percentUsed) ?? 0.5);
-  const sessionWindow = getQuotaWindow(quota, "window5h");
-  const weeklyWindow = getQuotaWindow(quota, "window7d") || getQuotaWindow(quota, "windowWeekly");
+  const sessionWindow = resolveQuotaWindowByName(quota, "session");
+  const weeklyWindow = resolveQuotaWindowByName(quota, "weekly");
   const sessionRemaining = clamp01(1 - (sessionWindow?.percentUsed ?? overallPercentUsed));
   const weeklyRemaining = clamp01(1 - (weeklyWindow?.percentUsed ?? overallPercentUsed));
   const sessionScore = scoreQuotaWindow(

open-sse/services/combo/targetResolution.ts

@@ -122,13 +122,6 @@ export interface ResolvedComboTargetPipeline {
   /** Session-stickiness result — the attempt loop reads `.messageHash` on success/failure. */
   sticky: ApplyStickinessResult;
   preScreenMap: Map<string, PreScreenResult>;
-  /**
-   * Idempotent release for the in-flight slot quota-share ordering reserved for
-   * its winner (#11371). Null unless the `quota-share` strategy ran. The host MUST
-   * invoke it when the request settles; this pipeline already releases it on any
-   * earlyResponse it produces after selection.
-   */
-  quotaShareRelease: (() => void) | null;
 }
 
 export type ResolveComboTargetPipelineResult =
@@ -401,11 +394,7 @@ async function orderByStrategy(
   initialOrderedTargets: ResolvedComboTarget[]
 ): Promise<
   | { earlyResponse: Response }
-  | {
-      orderedTargets: ResolvedComboTarget[];
-      autoUsedExplicitRouter: boolean;
-      quotaShareRelease: (() => void) | null;
-    }
+  | { orderedTargets: ResolvedComboTarget[]; autoUsedExplicitRouter: boolean }
 > {
   const { strategy, body, combo, settings, config, log } = deps;
   if (strategy === "auto") {
@@ -424,22 +413,17 @@ async function orderByStrategy(
     return {
       orderedTargets: autoResult.orderedTargets,
       autoUsedExplicitRouter: autoResult.autoUsedExplicitRouter,
-      quotaShareRelease: null,
     };
   }
-  const { orderedTargets, quotaShareRelease } = await applyStrategyOrdering(
-    strategy,
-    initialOrderedTargets,
-    {
-      combo,
-      config,
-      body,
-      log,
-      apiKeyAllowedConnections: deps.apiKeyAllowedConnections,
-      sessionKey: deps.relayOptions?.sessionId,
-    }
-  );
-  return { orderedTargets, autoUsedExplicitRouter: false, quotaShareRelease };
+  const orderedTargets = await applyStrategyOrdering(strategy, initialOrderedTargets, {
+    combo,
+    config,
+    body,
+    log,
+    apiKeyAllowedConnections: deps.apiKeyAllowedConnections,
+    sessionKey: deps.relayOptions?.sessionId,
+  });
+  return { orderedTargets, autoUsedExplicitRouter: false };
 }
 
 /**
@@ -730,15 +714,10 @@ export async function resolveComboTargetPipeline(
 
   const ordering = await orderByStrategy(deps, orderedTargets);
   if ("earlyResponse" in ordering) return ordering;
-  const { autoUsedExplicitRouter, quotaShareRelease } = ordering;
+  const { autoUsedExplicitRouter } = ordering;
 
   const continuity = await applyContinuityFilters(deps, ordering.orderedTargets);
-  if ("earlyResponse" in continuity) {
-    // #11371: selection already reserved the winner's in-flight slot; a hard
-    // filter exhausting the pool must not leak it.
-    quotaShareRelease?.();
-    return continuity;
-  }
+  if ("earlyResponse" in continuity) return continuity;
   orderedTargets = applyTaskAwareOrdering(deps, continuity.orderedTargets, autoUsedExplicitRouter);
   orderedTargets = await applyPromptCacheStage(
     deps,
@@ -762,6 +741,5 @@ export async function resolveComboTargetPipeline(
     getWeightedStepKeyForTarget,
     sticky: continuity.sticky,
     preScreenMap,
-    quotaShareRelease,
   };
 }

open-sse/services/compression/eval/executorModelClient.ts

@@ -16,9 +16,10 @@ export function createExecutorModelClient(
   credentials: ProviderCredentials,
   costPerKTokenOut?: number
 ): ModelClient {
-  const executor = getExecutor(provider);
   return {
     async complete(model: string, messages: ChatTurn[]): Promise<ModelCallResult> {
+      // #11220: getExecutor is async (lazy registry) — resolve per call.
+      const executor = await getExecutor(provider);
       const body = { model, messages, stream: false };
       const input: ExecuteInput = {
         model,

open-sse/services/genericQuotaFetcher.ts

@@ -150,16 +150,65 @@ export function convertUsageToQuotaInfo(usage: unknown): QuotaInfo | null {
 
   if (Object.keys(windows).length === 0) return null;
 
+  const normalized = normalizeQuotaWindows(windows);
+  Object.assign(windows, normalized);
+
   return {
     used: 0,
     total: 0,
     percentUsed: worstPercent,
     resetAt: worstResetAt,
     windows,
+    ...normalized,
     limitReached: worstPercent >= 1 - 1e-9,
   };
 }
 
+/**
+ * Map provider-native window keys to canonical structural windows so that
+ * reset-aware / reset-window scoring works without knowing every provider's
+ * naming convention.
+ *
+ *   - Claude: "session (5h)" → window5h, "weekly (7d)" → window7d
+ *   - Antigravity: worst per-model quota → window5h; worst *_weekly quota → window7d
+ */
+function normalizeQuotaWindows(
+  windows: Record<string, { percentUsed: number; resetAt: string | null }>
+): Record<string, { percentUsed: number; resetAt: string | null }> {
+  const normalized: Record<string, { percentUsed: number; resetAt: string | null }> = {};
+
+  // Claude-style explicit time windows.
+  if (windows["session (5h)"] && !normalized.window5h) {
+    normalized.window5h = windows["session (5h)"];
+  }
+  if (windows["weekly (7d)"] && !normalized.window7d) {
+    normalized.window7d = windows["weekly (7d)"];
+  }
+
+  // Antigravity-style per-model 5h windows: pick the worst (most used) model quota.
+  const modelWindows = Object.entries(windows).filter(
+    ([key]) =>
+      key !== "credits" &&
+      !key.endsWith("_weekly") &&
+      !key.startsWith("window") &&
+      !key.includes("(5h)") &&
+      !key.includes("(7d)")
+  );
+  if (modelWindows.length > 0 && !normalized.window5h) {
+    const worst = modelWindows.reduce((a, b) => (a[1].percentUsed > b[1].percentUsed ? a : b));
+    normalized.window5h = worst[1];
+  }
+
+  // Antigravity-style weekly family buckets: pick the worst *_weekly quota.
+  const weeklyWindows = Object.entries(windows).filter(([key]) => key.endsWith("_weekly"));
+  if (weeklyWindows.length > 0 && !normalized.window7d) {
+    const worst = weeklyWindows.reduce((a, b) => (a[1].percentUsed > b[1].percentUsed ? a : b));
+    normalized.window7d = worst[1];
+  }
+
+  return normalized;
+}
+
 /**
  * Fetch quota for a connection by delegating to the appropriate
  * provider-specific usage fetcher and reshaping its output into the

open-sse/services/quotaPreflight.ts

@@ -48,6 +48,16 @@ export interface QuotaInfo {
    * (e.g. "session", "weekly", "monthly").
    */
   windows?: Record<string, QuotaWindowInfo>;
+  /**
+   * Structural, canonical window snapshots used by reset-aware / reset-window
+   * scoring. Providers that expose time-based windows (5h, weekly, monthly)
+   * populate these in addition to the provider-native `windows` map so the
+   * scorer does not need to know every provider's key naming convention.
+   */
+  window5h?: QuotaWindowInfo;
+  window7d?: QuotaWindowInfo;
+  windowWeekly?: QuotaWindowInfo;
+  windowMonthly?: QuotaWindowInfo;
   /** True when the upstream usage endpoint explicitly reports exhausted quota. */
   limitReached?: boolean;
 }

open-sse/services/volcengineConsoleAutoLogin.ts

@@ -96,13 +96,6 @@ const ARK_CONSOLE_URL =
 /** Cookie names required for a valid console session (mirrors tokenExtractionConfig) */
 const REQUIRED_COOKIES = ["digest", "AccountID", "csrfToken", "userInfo"] as const;
 
-/** Exact-domain match for session cookies — substring checks would also accept
- * look-alike hosts (e.g. `volcengine.com.evil.test`). Playwright may report the
- * domain with or without a leading dot. */
-function isVolcengineCookieDomain(domain: string): boolean {
-  return domain === "volcengine.com" || domain.endsWith(".volcengine.com");
-}
-
 const DEFAULT_SESSION_TIMEOUT = 300_000;
 const SUBMIT_COOKIE_TIMEOUT = 90_000;
 const CAPTURE_POLL_INTERVAL = 1_000;
@@ -625,7 +618,7 @@ export class VolcengineConsoleAutoLoginService {
       for (const cookie of cookies as Array<{ name: string; domain: string; value: string }>) {
         if (
           REQUIRED_COOKIES.includes(cookie.name as (typeof REQUIRED_COOKIES)[number]) &&
-          isVolcengineCookieDomain(cookie.domain)
+          cookie.domain.includes("volcengine.com")
         ) {
           credentials[cookie.name] = cookie.value;
         }
@@ -770,7 +763,7 @@ export class VolcengineConsoleAutoLoginService {
           domain: string;
         }>;
         const present = REQUIRED_COOKIES.filter((name) =>
-          cookies.some((c) => c.name === name && isVolcengineCookieDomain(c.domain))
+          cookies.some((c) => c.name === name && c.domain.includes("volcengine.com"))
         );
         parts.push(
           `cookies=[${present.join(",") || "none of digest/AccountID/csrfToken/userInfo"}]`

package.json

@@ -1,7 +1,7 @@
 {
   "name": "omniroute",
   "version": "3.8.51",
-  "description": "Unified AI router with 353 providers, RTK+Caveman compression, auto fallback, MCP/A2A, desktop, PWA, and OpenAI-compatible APIs.",
+  "description": "Unified AI router with 350 providers, RTK+Caveman compression, auto fallback, MCP/A2A, desktop, PWA, and OpenAI-compatible APIs.",
   "type": "module",
   "bin": {
     "omniroute": "bin/omniroute.mjs",
@@ -261,7 +261,7 @@
     "alibaba:sync-allowlist": "node --import tsx/esm scripts/ops/sync-alibaba-allowlist.mjs"
   },
   "dependencies": {
-    "@aws-sdk/client-bedrock-runtime": "^3.1112.0",
+    "@aws-sdk/client-bedrock-runtime": "^3.1116.0",
     "@dnd-kit/core": "^6.3.1",
     "@dnd-kit/sortable": "^10.0.0",
     "@dnd-kit/utilities": "^3.2.2",
@@ -280,29 +280,29 @@
     "commander": "^15.0.0",
     "cron-parser": "^5.10.0",
     "csv-stringify": "^6.8.3",
-    "dompurify": "^3.4.13",
+    "dompurify": "^3.4.14",
     "express": "^5.2.1",
     "fetch-socks": "^1.3.3",
     "fflate": "^0.8.3",
-    "fumadocs-core": "^16.14.4",
-    "fumadocs-ui": "^16.14.4",
+    "fumadocs-core": "^16.15.1",
+    "fumadocs-ui": "^16.15.1",
     "http-proxy-middleware": "^4.0.0",
     "https-proxy-agent": "^9.0.0",
     "ink": "^7.0.3",
     "ink-spinner": "^5.0.0",
     "ink-text-input": "^6.0.0",
     "ioredis": "^5.10.1",
-    "jose": "^6.2.9",
+    "jose": "^6.2.10",
     "js-yaml": "^5.3.0",
     "jsonc-parser": "^3.3.1",
     "lowdb": "^7.0.1",
-    "lucide-react": "^1.21.0",
-    "marked": "^18.0.9",
+    "lucide-react": "^1.33.0",
+    "marked": "^18.0.10",
     "marked-terminal": "^7.3.0",
     "material-symbols": "^0.46.0",
-    "mermaid": "^11.15.0",
+    "mermaid": "^11.17.0",
     "monaco-editor": "^0.56.0",
-    "next": "16.3.1",
+    "next": "16.3.2",
     "next-intl": "^4.13.7",
     "next-themes": "^0.4.6",
     "node-machine-id": "^1.1.12",
@@ -332,7 +332,7 @@
     "turndown-plugin-gfm": "1.0.2",
     "undici": "^8.10.0",
     "update-notifier": "^7.3.1",
-    "uuid": "^14.0.0",
+    "uuid": "^14.0.2",
     "ws": "^8.21.3",
     "xxhash-wasm": "^1.1.0",
     "yazl": "^3.3.1",
@@ -348,11 +348,12 @@
     "onnxruntime-node": "1.24.3",
     "sqlite-vec": "^0.1.9",
     "tls-client-node": "^0.2.0",
-    "wreq-js": "^3.0.0"
+    "wreq-js": "^3.1.0"
   },
   "devDependencies": {
     "@axe-core/playwright": "^4.13.0",
     "@cyclonedx/cyclonedx-npm": "6.0.1",
+    "@eslint/compat": "^2.1.0",
     "@playwright/test": "^1.62.1",
     "@size-limit/file": "^13.0.3",
     "@stryker-mutator/core": "^10.0.0",
@@ -363,24 +364,25 @@
     "@testing-library/react": "^16.3.2",
     "@testing-library/user-event": "^14.6.6",
     "@types/better-sqlite3": "^9.6.0",
-    "@types/bun": "latest",
+    "@types/bun": "1.4.0",
     "@types/node": "^26.2.0",
     "@types/react": "^19.2.18",
     "@types/react-dom": "^19.2.4",
     "@types/safe-regex": "^1.1.6",
     "@types/ws": "^8.18.0",
-    "@vitejs/plugin-react": "^6.0.5",
-    "bun": "1.3.14",
+    "@vitejs/plugin-react": "^6.1.0",
+    "bun": "1.4.0",
     "c8": "^12.0.0",
     "concurrently": "^10.0.5",
     "cross-env": "^10.1.0",
     "ctrf": "^0.3.0",
     "dpdm": "^4.3.0",
-    "eslint": "^9.39.4",
-    "eslint-config-next": "16.3.1",
+    "eslint": "^10.9.0",
+    "eslint-config-next": "16.3.2",
     "eslint-plugin-sonarjs": "^4.1.0",
+    "espree": "^11.2.0",
     "fast-check": "^4.8.0",
-    "fumadocs-mdx": "^15.2.3",
+    "fumadocs-mdx": "^15.3.1",
     "glob": "^13.0.6",
     "httpyac": "^6.16.7",
     "husky": "^9.1.7",
@@ -392,7 +394,7 @@
     "lint-staged": "^17.3.0",
     "lockfile-lint": "^5.0.1",
     "node-loader": "^2.1.0",
-    "opencode-ai": "1.18.18",
+    "opencode-ai": "1.18.21",
     "playwright-ctrf-json-reporter": "^0.0.29",
     "prettier": "^3.9.6",
     "promptfoo": "^0.122.0",
@@ -401,7 +403,7 @@
     "type-coverage": "^2.30.1",
     "typescript": "^6.0.3",
     "typescript-eslint": "^8.67.0",
-    "vitest": "^4.1.7",
+    "vitest": "^4.1.11",
     "wait-on": "^9.1.0",
     "wtfnode": "^0.10.1"
   },
@@ -435,12 +437,13 @@
   },
   "overrides": {
     "onnxruntime-node": "1.24.3",
+    "eslint-plugin-react-hooks": "7.0.1",
     "fast-xml-parser": "^5.10.1",
     "sharp": "^0.35.3",
     "postcss": "^8.5.18",
     "ip-address": "^10.3.1",
     "qs": "^6.15.2",
-    "uuid": "^14.0.0",
+    "uuid": "^14.0.2",
     "form-data": "^4.0.6",
     "vite": "^8.0.16",
     "protobufjs": "^7.6.5",
@@ -468,10 +471,6 @@
     },
     "socket.io-parser": "^4.2.7",
     "tar": "^7.5.21",
-    "brace-expansion": "^5.0.9",
-    "minimatch": {
-      "brace-expansion": "^1.1.18"
-    },
     "libxmljs2": {
       "minimatch": {
         "brace-expansion": "^2.1.4"
@@ -496,7 +495,7 @@
     },
     "nanoid": "^3.3.17",
     "monaco-editor": {
-      "dompurify": "^3.4.13"
+      "dompurify": "^3.4.14"
     }
   }
 }

scripts/build/afterpack-copy-node-modules.mjs

@@ -0,0 +1,34 @@
+import { cpSync, existsSync, readdirSync, rmSync } from "node:fs";
+import { join } from "node:path";
+
+// electron-builder >=26 injects an "!**/node_modules/**" ignore into every
+// extraResources/extraFiles pattern list (app-builder-lib/out/fileMatcher.js),
+// and that ignore cannot be overridden by any later positive filter pattern —
+// verified empirically with a minimal fixture on 26.15.3. The standalone
+// server resolves better-sqlite3 (and other runtime deps) from the *primary*
+// node_modules at resources/app/node_modules (see
+// prepare-electron-standalone.mjs: "Verify better-sqlite3 Node-API prebuilds in
+// the primary node_modules"), so without this hook the packaged desktop app silently loses its native
+// SQLite driver and falls back to sql.js — the exact regression guarded by
+// issue #7592's cold-restart smoke check.
+export default async function afterPack(context) {
+  const stagingNodeModules = join(
+    context.packager.projectDir,
+    "..",
+    ".build",
+    "electron-standalone",
+    "node_modules"
+  );
+  const destNodeModules = join(context.appOutDir, "resources", "app", "node_modules");
+
+  if (!existsSync(stagingNodeModules)) {
+    console.warn(`[afterpack] no staged node_modules at ${stagingNodeModules} — skipping restore`);
+    return;
+  }
+
+  rmSync(destNodeModules, { recursive: true, force: true });
+  cpSync(stagingNodeModules, destNodeModules, { recursive: true });
+  console.log(
+    `[afterpack] restored ${readdirSync(destNodeModules).length} runtime module(s) into resources/app/node_modules`
+  );
+}

scripts/build/optionalPackStaging.mjs

@@ -89,16 +89,20 @@ function moveTree(src, dest) {
   return { removedFiles: files, removedBytes: bytes };
 }
 
-function tarPack(packOutDir, tarballPath) {
+export function tarPack(packOutDir, tarballPath) {
   // bsdtar ships with macOS, Linux images, and Windows runners (System32\tar.exe).
+  // GNU tar (common in Git-for-Windows environments) treats `C:\...` in `-f` as a
+  // remote rsh target ("Cannot connect to C:"), so always pass a bare filename
+  // and point cwd at the tarball directory instead.
   const result = spawnSync(
     process.platform === "win32" ? "tar.exe" : "tar",
-    ["-czf", tarballPath, "-C", packOutDir, "node_modules"],
-    { stdio: "pipe" }
+    ["-czf", path.basename(tarballPath), "-C", packOutDir, "node_modules"],
+    { stdio: "pipe", cwd: path.dirname(tarballPath) }
   );
   if (result.status !== 0) {
     throw new Error(
-      `optional-pack tar failed for ${path.basename(tarballPath)} (exit ${result.status})`
+      `optional-pack tar failed for ${path.basename(tarballPath)} (exit ${result.status})` +
+        `: ${(result.stderr || result.stdout || "").toString().slice(-600)}`
     );
   }
 }

scripts/dev/smoke-electron-packaged.mjs

@@ -360,7 +360,7 @@ function isInsideDir(parentDir, candidateDir) {
   return candidate === parent || candidate.startsWith(parent + sep);
 }
 
-async function ensureSmokeEnvDirs(smokeEnv, dataDir) {
+export async function ensureSmokeEnvDirs(smokeEnv, dataDir) {
   const dirNames = [
     "DATA_DIR",
     "HOME",
@@ -388,6 +388,16 @@ async function ensureSmokeEnvDirs(smokeEnv, dataDir) {
       dirs.push(join(smokeEnv.APPDATA, subdir));
     }
   }
+  // Electron resolves the Roaming profile from %USERPROFILE%\AppData\Roaming
+  // (USERPROFILE takes precedence over the APPDATA env var) and the path
+  // service throws — rather than creates — when that directory is missing,
+  // which makes requestSingleInstanceLock() return false and the app exit(0)
+  // before app.whenReady(). Pre-create the derived tree as well.
+  if (platform() === "win32" && smokeEnv.USERPROFILE) {
+    for (const subdir of ["omniroute-desktop", "OmniRoute", "omniroute"]) {
+      dirs.push(join(smokeEnv.USERPROFILE, "AppData", "Roaming", subdir));
+    }
+  }
 
   await Promise.all(dirs.map((dir) => mkdir(dir, { recursive: true })));
 }
@@ -506,7 +516,14 @@ async function waitForReady({ logs, smokeUrl, timeoutMs, settleMs, exitState })
  * by the single-launch path and the cold-restart (two-launch) path so both
  * exercise identical spawn/readiness/shutdown behavior.
  */
-async function launchAndCollectLogs({ appExecutable, smokeUrl, dataDir, timeoutMs, settleMs, streamLogs }) {
+async function launchAndCollectLogs({
+  appExecutable,
+  smokeUrl,
+  dataDir,
+  timeoutMs,
+  settleMs,
+  streamLogs,
+}) {
   const smokeEnv = buildSmokeEnv({ dataDir });
   await assertPortIsFree(smokeUrl);
   await ensureSmokeEnvDirs(smokeEnv, dataDir);
@@ -568,7 +585,14 @@ async function main() {
     !process.env.ELECTRON_SMOKE_DATA_DIR && process.env.ELECTRON_SMOKE_KEEP_DATA !== "1";
 
   try {
-    await launchAndCollectLogs({ appExecutable, smokeUrl, dataDir, timeoutMs, settleMs, streamLogs });
+    await launchAndCollectLogs({
+      appExecutable,
+      smokeUrl,
+      dataDir,
+      timeoutMs,
+      settleMs,
+      streamLogs,
+    });
 
     if (!coldRestart) return;
 

src/app/(dashboard)/dashboard/memory/components/EmbeddingSourceSelector.tsx

@@ -89,22 +89,46 @@ export default function EmbeddingSourceSelector({ settings, providers, onSave, s
               {t("embedding.noRemoteProviders")}
             </p>
           ) : (
-            <select
-              value={currentProviderModel}
-              onChange={(e) => handleProviderModelChange(e.target.value)}
-              disabled={saving}
-              data-testid="embedding-provider-model-select"
-              className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm focus:outline-none focus:ring-1 focus:ring-violet-500"
-            >
-              <option value="">{t("embedding.selectProviderModel")}</option>
-              {remoteProviders.map((p) =>
-                p.models.map((m) => (
-                  <option key={m.id} value={m.id}>
-                    {m.name} ({m.dimensions ? `${m.dimensions}d` : "?"})
-                  </option>
-                ))
-              )}
-            </select>
+            <>
+              <select
+                value={
+                  remoteProviders.some((p) => p.models.some((m) => m.id === currentProviderModel))
+                    ? currentProviderModel
+                    : ""
+                }
+                onChange={(e) => handleProviderModelChange(e.target.value)}
+                disabled={saving}
+                data-testid="embedding-provider-model-select"
+                className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm focus:outline-none focus:ring-1 focus:ring-violet-500"
+              >
+                <option value="">{t("embedding.selectProviderModel")}</option>
+                {remoteProviders.map((p) =>
+                  p.models.length > 0 ? (
+                    p.models.map((m) => (
+                      <option key={m.id} value={m.id}>
+                        {m.name} ({m.dimensions ? `${m.dimensions}d` : "?"})
+                      </option>
+                    ))
+                  ) : (
+                    <optgroup key={p.provider} label={p.provider}>
+                      <option value="">{`— ${p.provider} (no curated models)`}</option>
+                    </optgroup>
+                  )
+                )}
+              </select>
+              {/* Free-text override: the runtime accepts any configured provider's
+                OpenAI-compatible model id, including ones without a curated
+                registry entry (e.g. groq/, mistral/, cf/...). */}
+              <input
+                type="text"
+                value={currentProviderModel}
+                onChange={(e) => handleProviderModelChange(e.target.value)}
+                disabled={saving}
+                placeholder="provider/model — e.g. mistral/mistral-embed"
+                data-testid="embedding-provider-model-input"
+                className="w-full mt-2 px-3 py-2 rounded-lg bg-background border border-border text-sm font-mono focus:outline-none focus:ring-1 focus:ring-violet-500"
+              />
+            </>
           )}
           <CustomEmbeddingEndpointFields settings={settings} onSave={onSave} saving={saving} />
         </div>

src/app/(dashboard)/dashboard/memory/components/RerankConfigCard.tsx

@@ -43,11 +43,7 @@ export default function RerankConfigCard({ settings, providers, onSave, saving }
           }}
           disabled={saving || (!rerankEnabled && !hasProvider)}
           aria-disabled={saving || (!rerankEnabled && !hasProvider)}
-          title={
-            !rerankEnabled && !hasProvider
-              ? t("rerank.noProviderWithKey")
-              : undefined
-          }
+          title={!rerankEnabled && !hasProvider ? t("rerank.noProviderWithKey") : undefined}
           role="switch"
           aria-checked={rerankEnabled}
           className={`relative w-11 h-6 rounded-full transition-colors shrink-0 disabled:opacity-50 disabled:cursor-not-allowed ${
@@ -83,22 +79,45 @@ export default function RerankConfigCard({ settings, providers, onSave, saving }
                 {t("rerank.noProviderWithKey")}
               </p>
             ) : (
-              <select
-                value={rerankProviderModel}
-                onChange={(e) => handleProviderModelChange(e.target.value)}
-                disabled={saving}
-                data-testid="rerank-provider-model-select"
-                className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm focus:outline-none focus:ring-1 focus:ring-violet-500"
-              >
-                <option value="">{t("rerank.selectProviderModel")}</option>
-                {rerankProviders.map((p) =>
-                  p.models.map((m) => (
-                    <option key={m.id} value={m.id}>
-                      {m.name}
-                    </option>
-                  )),
-                )}
-              </select>
+              <>
+                <select
+                  value={
+                    rerankProviders.some((p) => p.models.some((m) => m.id === rerankProviderModel))
+                      ? rerankProviderModel
+                      : ""
+                  }
+                  onChange={(e) => handleProviderModelChange(e.target.value)}
+                  disabled={saving}
+                  data-testid="rerank-provider-model-select"
+                  className="w-full px-3 py-2 rounded-lg bg-background border border-border text-sm focus:outline-none focus:ring-1 focus:ring-violet-500"
+                >
+                  <option value="">{t("rerank.selectProviderModel")}</option>
+                  {rerankProviders.map((p) =>
+                    p.models.length > 0 ? (
+                      p.models.map((m) => (
+                        <option key={m.id} value={m.id}>
+                          {m.name}
+                        </option>
+                      ))
+                    ) : (
+                      <optgroup key={p.provider} label={p.provider}>
+                        <option value="">{`— ${p.provider} (no curated models)`}</option>
+                      </optgroup>
+                    )
+                  )}
+                </select>
+                {/* Free-text override: any configured provider's Cohere-compatible
+                  model id is accepted by the runtime even without a curated entry. */}
+                <input
+                  type="text"
+                  value={rerankProviderModel}
+                  onChange={(e) => handleProviderModelChange(e.target.value)}
+                  disabled={saving}
+                  placeholder="provider/model — e.g. groq/my-reranker"
+                  data-testid="rerank-provider-model-input"
+                  className="w-full mt-2 px-3 py-2 rounded-lg bg-background border border-border text-sm font-mono focus:outline-none focus:ring-1 focus:ring-violet-500"
+                />
+              </>
             )}
           </div>
         </>

src/app/(dashboard)/dashboard/memory/components/tabs/EngineTab.tsx

@@ -16,6 +16,7 @@ export default function EngineTab() {
   const { status, isLoading: statusLoading } = useEngineStatus();
   const { settings, save: saveSettings, isLoading: settingsLoading } = useMemorySettings();
   const [providers, setProviders] = useState<EmbeddingProviderListing[]>([]);
+  const [rerankProviders, setRerankProviders] = useState<EmbeddingProviderListing[]>([]);
   const [saving, setSaving] = useState(false);
   const [reindexing, setReindexing] = useState(false);
   const [reindexMsg, setReindexMsg] = useState("");
@@ -32,6 +33,14 @@ export default function EngineTab() {
         if (!cancelled && data?.providers) setProviders(data.providers);
       })
       .catch(() => {});
+    // Rerank has its own curated registry — the embedding listing does not
+    // include rerank-only providers (cohere rerank SKUs, siliconflow, ...).
+    fetch("/api/memory/rerank-providers")
+      .then((r) => (r.ok ? r.json() : null))
+      .then((data) => {
+        if (!cancelled && data?.providers) setRerankProviders(data.providers);
+      })
+      .catch(() => {});
     return () => {
       cancelled = true;
     };
@@ -139,7 +148,7 @@ export default function EngineTab() {
             </h3>
             <RerankConfigCard
               settings={settings}
-              providers={providers}
+              providers={rerankProviders}
               onSave={handleSaveSettings}
               saving={saving}
             />