fix(antigravity): map Gemini 3.7 Flash tiers to upstream tiered endpoint model (#10341) (#10882)

Merged — clean single-commit cherry-pick extracted from #10879's genuinely new content (see PR body for the full extraction rationale). typecheck/file-size/changelog gates clean, 11/11 tests passing.

changelog.d/fixes/10882-antigravity-gemini37-flash-tiers.md

@@ -0,0 +1 @@
+- **fix(antigravity):** map Gemini 3.7 Flash tier ids (`gemini-3.7-flash-high/medium/low`, bare `gemini-3.7-flash`) to the upstream `gemini-3.7-flash-tiered` model id Google's Cloud Code endpoint expects, and configure per-tier thinking budgets ([#10882](https://github.com/diegosouzapw/OmniRoute/pull/10882)) — thanks @adevwithpurpose

open-sse/config/agyModels.ts

@@ -41,6 +41,15 @@ export const AGY_PUBLIC_MODELS = Object.freeze([
     supportsVision: true,
     toolCalling: true,
   },
+  {
+    id: "gemini-3.7-flash-tiered",
+    name: "Gemini 3.7 Flash (Tiered)",
+    contextLength: 1048576,
+    maxOutputTokens: 65536,
+    supportsReasoning: true,
+    supportsVision: true,
+    toolCalling: true,
+  },
   // Gemini 3.1 Pro
   {
     id: "gemini-pro-agent",

open-sse/config/antigravityModelAliases.ts

@@ -29,6 +29,15 @@ export const ANTIGRAVITY_PUBLIC_MODELS = Object.freeze([
     supportsVision: true,
     toolCalling: true,
   },
+  {
+    id: "gemini-3.7-flash-tiered",
+    name: "Gemini 3.7 Flash (Tiered)",
+    contextLength: 1048576,
+    maxOutputTokens: 65536,
+    supportsReasoning: true,
+    supportsVision: true,
+    toolCalling: true,
+  },
   // Gemini 3.1 Pro budget tiers. Live streamGenerateContent validation uses
   // `gemini-pro-agent` for High; the separately advertised `gemini-3.1-pro-high`
   // discovery slot currently returns HTTP 400 and is intentionally not public.
@@ -91,6 +100,13 @@ export const ANTIGRAVITY_PUBLIC_MODELS = Object.freeze([
 ]);
 
 export const ANTIGRAVITY_MODEL_ALIASES = Object.freeze({
+  // Gemini 3.7 Flash tiers map to the upstream tiered endpoint model; the thinking
+  // budget is steered via generationConfig.thinkingConfig.thinkingBudget.
+  "gemini-3.7-flash": "gemini-3.7-flash-tiered",
+  "gemini-3.7-flash-high": "gemini-3.7-flash-tiered",
+  "gemini-3.7-flash-medium": "gemini-3.7-flash-tiered",
+  "gemini-3.7-flash-low": "gemini-3.7-flash-tiered",
+  "gpt-oss-120b": "gpt-oss-120b-medium",
   // gemini-3.1-pro-low is not aliased: the upstream accepts it verbatim.
   // gemini-3.1-pro-high: the discovery slot returns HTTP 400 on v1internal;
   // the live upstream id is gemini-pro-agent (see ANTIGRAVITY_PUBLIC_MODELS).

src/shared/constants/modelSpecs.ts

@@ -175,12 +175,54 @@ export const MODEL_SPECS: Record<string, ModelSpec> = {
   },
 
   // ── Gemini 3.7 Flash (current Antigravity/AGY live tiers) ─────────
-  // The model id itself selects the upstream 10k/4k/1k reasoning tier. Antigravity
-  // still rejects client-supplied thinking parameters, so keep the explicit-parameter
-  // capability aligned with the existing Gemini Flash tier ids.
-  "gemini-3.7-flash-high": { ...GEMINI_35_FLASH_MODEL_SPEC },
-  "gemini-3.7-flash-medium": { ...GEMINI_35_FLASH_MODEL_SPEC },
-  "gemini-3.7-flash-low": { ...GEMINI_35_FLASH_MODEL_SPEC },
+  // The tier suffix configures the thinking budget passed to the upstream
+  // gemini-3.7-flash-tiered backend (high: 24.5k, medium: 8k, low: 1k).
+  "gemini-3.7-flash-high": {
+    maxOutputTokens: 65536,
+    contextWindow: 1048576,
+    defaultThinkingBudget: 24576,
+    thinkingBudgetCap: 24576,
+    supportsThinking: true,
+    supportsTools: true,
+    supportsVision: true,
+  },
+  "gemini-3.7-flash-medium": {
+    maxOutputTokens: 65536,
+    contextWindow: 1048576,
+    defaultThinkingBudget: 8192,
+    thinkingBudgetCap: 24576,
+    supportsThinking: true,
+    supportsTools: true,
+    supportsVision: true,
+  },
+  "gemini-3.7-flash-low": {
+    maxOutputTokens: 65536,
+    contextWindow: 1048576,
+    defaultThinkingBudget: 1024,
+    thinkingBudgetCap: 24576,
+    supportsThinking: true,
+    supportsTools: true,
+    supportsVision: true,
+  },
+  "gemini-3.7-flash": {
+    maxOutputTokens: 65536,
+    contextWindow: 1048576,
+    defaultThinkingBudget: 8192,
+    thinkingBudgetCap: 24576,
+    supportsThinking: true,
+    supportsTools: true,
+    supportsVision: true,
+    aliases: ["gemini-3.7-flash-tiered"],
+  },
+  "gemini-3.7-flash-tiered": {
+    maxOutputTokens: 65536,
+    contextWindow: 1048576,
+    defaultThinkingBudget: 8192,
+    thinkingBudgetCap: 24576,
+    supportsThinking: true,
+    supportsTools: true,
+    supportsVision: true,
+  },
 
   // Provider-neutral compatibility for providers that still serve Gemini 3.6.
   // Antigravity/AGY availability is governed by their own provider catalogs and

tests/unit/agy-gemini-3696-tier-passthrough.test.ts

@@ -14,10 +14,13 @@ test("(#3696) resolveAntigravityModelId passes gemini-3.1-pro-low through unchan
   assert.equal(resolveAntigravityModelId("gemini-3.1-pro-low"), "gemini-3.1-pro-low");
 });
 
-test("(#3696) no two ANTIGRAVITY_PUBLIC_MODELS entries resolve to the same upstream id", () => {
+test("(#3696) non-tiered ANTIGRAVITY_PUBLIC_MODELS entries resolve to distinct upstream ids", () => {
   const seen = new Map<string, string>();
   const collisions: string[] = [];
   for (const model of ANTIGRAVITY_PUBLIC_MODELS) {
+    // Gemini 3.7 Flash tiers intentionally share the upstream `gemini-3.7-flash-tiered`
+    // endpoint with different reasoning token budgets.
+    if (model.id.startsWith("gemini-3.7-flash")) continue;
     const upstream = resolveAntigravityModelId(model.id);
     if (seen.has(upstream)) {
       collisions.push(`${model.id} and ${seen.get(upstream)} both resolve to "${upstream}"`);

tests/unit/antigravity-model-aliases.test.ts

@@ -57,12 +57,11 @@ test("toClientAntigravityQuotaModelId preserves upstream Gemini Flash bucket IDs
 test("resolveAntigravityModelId maps the documented Antigravity aliases to upstream IDs", () => {
   assert.equal(resolveAntigravityModelId("gemini-3-pro-image-preview"), "gemini-3-pro-image");
   for (const [modelId] of EXPECTED_FLASH_TIERS) {
-    // Only the collapsed gemini-3.7-flash id is aliased to the live upstream
-    // gemini-3.7-flash-tiered id; the suffixed gemini-3.7-flash-high/medium tier ids
-    // (like the 3.6/3.5 tiers) have no alias entry and pass through verbatim.
-    const expected = modelId === "gemini-3.7-flash" ? "gemini-3.7-flash-tiered" : modelId;
-    assert.equal(resolveAntigravityModelId(modelId), expected);
+    assert.equal(resolveAntigravityModelId(modelId), "gemini-3.7-flash-tiered");
   }
+  assert.equal(resolveAntigravityModelId("gemini-3.7-flash"), "gemini-3.7-flash-tiered");
+  assert.equal(resolveAntigravityModelId("gemini-3.7-flash-tiered"), "gemini-3.7-flash-tiered");
+  assert.equal(resolveAntigravityModelId("gpt-oss-120b"), "gpt-oss-120b-medium");
   assert.equal(resolveAntigravityModelId("gemini-claude-sonnet-4-5"), "claude-sonnet-4-6");
   assert.equal(resolveAntigravityModelId("gemini-claude-sonnet-4-5-thinking"), "claude-sonnet-4-6");
   assert.equal(