feat(codex): accept parenthesized GPT-5.6 effort overrides (#9208)

Validated in local merge-train (devbox-vm-06-dev002) @ combined-tip (FAST gates — only pre-existing audit.test.ts flake).

changelog.d/features/9208-codex-parenthesized-reasoning.md

@@ -0,0 +1 @@
+- **feat(codex):** accept parenthesized GPT-5.6 reasoning overrides. (thanks @seakleangnhak)

open-sse/executors/codex.ts

@@ -48,6 +48,12 @@ export {
   getCodexDualWindowCooldownMs,
 } from "./codex/quota.ts";
 import { isCodexFreePlan, normalizeCodexTools } from "./codex/tools.ts";
+import {
+  CODEX_EFFORT_ORDER as EFFORT_ORDER,
+  GPT_5_6_ULTRA_ALIAS_MODELS,
+  splitCodexReasoningSuffix,
+  type CodexEffortLevel as EffortLevel,
+} from "./codex/reasoningSuffix.ts";
 // Re-exported for external importers (tests + provider services).
 export { isCodexFreePlan, normalizeCodexTools } from "./codex/tools.ts";
 
@@ -117,12 +123,6 @@ function codexWebSocketUnavailableResponse(): Response {
 // Ref: sub2api PR #1129 (feat(openai): split codex spark rate limiting from codex)
 export { getCodexModelScope, getCodexRateLimitKey, type CodexQuotaScope };
 
-// Ordered list of effort levels from lowest to highest
-const EFFORT_ORDER = ["none", "low", "medium", "high", "xhigh", "max", "ultra"] as const;
-type EffortLevel = (typeof EFFORT_ORDER)[number];
-const STANDARD_EFFORT_SUFFIXES = ["none", "low", "medium", "high", "xhigh"] as const;
-const GPT_5_6_MAX_ALIAS_MODELS = new Set(["gpt-5.6-sol", "gpt-5.6-terra", "gpt-5.6-luna"]);
-const GPT_5_6_ULTRA_ALIAS_MODELS = new Set(["gpt-5.6-sol", "gpt-5.6-terra"]);
 const CODEX_FAST_WIRE_VALUE = "priority";
 const CODEX_RESPONSES_WS_URL = "wss://chatgpt.com/backend-api/codex/responses";
 const CODEX_RESPONSES_LITE_HEADER = "x-openai-internal-codex-responses-lite";
@@ -185,32 +185,6 @@ function enforceCodexResponsesLiteParallelToolCalls(
   return { ...body, parallel_tool_calls: false };
 }
 
-function splitCodexReasoningSuffix(model: unknown): {
-  baseModel: string;
-  effort: EffortLevel | null;
-} {
-  const modelId = typeof model === "string" ? model : "";
-  const gpt56AliasMatch = /^(gpt-5\.6-(?:sol|terra|luna))-(max|ultra)$/.exec(modelId);
-  if (gpt56AliasMatch) {
-    const [, baseModel, alias] = gpt56AliasMatch;
-    const supportedModels =
-      alias === "ultra" ? GPT_5_6_ULTRA_ALIAS_MODELS : GPT_5_6_MAX_ALIAS_MODELS;
-    if (supportedModels.has(baseModel)) {
-      return { baseModel, effort: alias as EffortLevel };
-    }
-  }
-
-  for (const level of STANDARD_EFFORT_SUFFIXES) {
-    if (modelId.endsWith(`-${level}`)) {
-      return {
-        baseModel: modelId.slice(0, -`-${level}`.length),
-        effort: level,
-      };
-    }
-  }
-  return { baseModel: modelId, effort: null };
-}
-
 export function getCodexUpstreamModel(model: unknown): string {
   return splitCodexReasoningSuffix(model).baseModel;
 }

open-sse/executors/codex/reasoningSuffix.ts

@@ -0,0 +1,41 @@
+export const CODEX_EFFORT_ORDER = [
+  "none",
+  "low",
+  "medium",
+  "high",
+  "xhigh",
+  "max",
+  "ultra",
+] as const;
+export type CodexEffortLevel = (typeof CODEX_EFFORT_ORDER)[number];
+export const GPT_5_6_MAX_ALIAS_MODELS = new Set(["gpt-5.6-sol", "gpt-5.6-terra", "gpt-5.6-luna"]);
+export const GPT_5_6_ULTRA_ALIAS_MODELS = new Set(["gpt-5.6-sol", "gpt-5.6-terra"]);
+
+export function splitCodexReasoningSuffix(model: unknown): {
+  baseModel: string;
+  effort: CodexEffortLevel | null;
+} {
+  const modelId = typeof model === "string" ? model : "";
+  const gpt56Match = /^(gpt-5\.6-(?:sol|terra|luna))(?:-(max|ultra)|\((max|ultra)\))$/.exec(
+    modelId
+  );
+  if (gpt56Match) {
+    const [, baseModel, hyphenEffort, parenthesizedEffort] = gpt56Match;
+    const effort = hyphenEffort ?? parenthesizedEffort;
+    const supportedModels = parenthesizedEffort
+      ? GPT_5_6_MAX_ALIAS_MODELS
+      : effort === "ultra"
+        ? GPT_5_6_ULTRA_ALIAS_MODELS
+        : GPT_5_6_MAX_ALIAS_MODELS;
+    if (supportedModels.has(baseModel)) {
+      return { baseModel, effort: effort as CodexEffortLevel };
+    }
+  }
+
+  for (const effort of ["none", "low", "medium", "high", "xhigh"] as const) {
+    if (modelId.endsWith(`-${effort}`)) {
+      return { baseModel: modelId.slice(0, -`-${effort}`.length), effort };
+    }
+  }
+  return { baseModel: modelId, effort: null };
+}

tests/unit/executor-codex-gpt56.test.ts

@@ -82,6 +82,32 @@ test("CodexExecutor.transformRequest clamps Luna ultra requests to its max effor
   assert.equal(result.reasoning.effort, "max");
 });
 
+test("CodexExecutor.transformRequest accepts parenthesized GPT-5.6 effort overrides", () => {
+  const executor = new CodexExecutor();
+  const cases = [
+    { model: "gpt-5.6-sol(ultra)", expectedModel: "gpt-5.6-sol", expectedEffort: "max" },
+    { model: "gpt-5.6-terra(max)", expectedModel: "gpt-5.6-terra", expectedEffort: "max" },
+    { model: "gpt-5.6-luna(ultra)", expectedModel: "gpt-5.6-luna", expectedEffort: "max" },
+  ];
+
+  for (const { model, expectedModel, expectedEffort } of cases) {
+    const result = executor.transformRequest(
+      model,
+      {
+        model,
+        input: [],
+        reasoning: { effort: "low", summary: "detailed" },
+      },
+      false,
+      { requestEndpointPath: "/responses" }
+    );
+
+    assert.equal(result.model, expectedModel);
+    assert.equal(result.reasoning.effort, expectedEffort);
+    assert.equal(result.reasoning.summary, "detailed");
+  }
+});
+
 test("CodexExecutor.execute disables parallel tool calls for Responses Lite markers", async () => {
   const executor = new CodexExecutor();
   const originalFetch = globalThis.fetch;