fix(combo): advance to next model on 400 'model not supported' (#5249)

A 400 classified as MODEL_CAPACITY (e.g. 'requested model is not supported') hit the #2101 anti-loop stop-branch and halted the combo instead of advancing. Drop the MODEL_CAPACITY trigger from that stop condition so model-specific 400s fall through to the next combo target; genuinely body-specific 400s (malformed/invalid/bad-request substrings) still stop. Validated: combo-strategies suite 16/16 (incl. the new regression test) green on current release tip, typecheck clean. Stale pre-merge CI was from an older base.

Co-authored-by: Chewji9875 <Chewji9875@users.noreply.github.com>

open-sse/services/combo.ts

@@ -17,7 +17,6 @@ import {
   recordProviderFailure,
   selectLockoutCooldownMs,
 } from "./accountFallback.ts";
-import { RateLimitReason } from "../config/constants.ts";
 import { errorResponse, unavailableResponse } from "../utils/error.ts";
 import {
   recordComboIntent,
@@ -2345,8 +2344,7 @@ export async function handleComboChat({
             fallbackResult.shouldFallback &&
             !isContextOverflow400(errorText) &&
             !isParamValidation400(errorText) &&
-            (fallbackResult.reason === RateLimitReason.MODEL_CAPACITY ||
-              errorText.toLowerCase().includes("context") ||
+            (errorText.toLowerCase().includes("context") ||
               errorText.toLowerCase().includes("prompt") ||
               errorText.toLowerCase().includes("token") ||
               errorText.toLowerCase().includes("malformed") ||

tests/unit/combo-strategies.test.ts

@@ -708,3 +708,39 @@ test("reset-aware strategy scores provider-specific weekly windows when availabl
 
   assert.equal(await selectedConnectionFor(combo), soon);
 });
+
+test("priority combo advances to next model when first returns 400 'model not supported'", async () => {
+  const name = `model-not-supported-${randomUUID()}`;
+  const combo = await combosDb.createCombo({
+    name,
+    strategy: "priority",
+    models: ["openai/gpt-4", "openai/gpt-3.5-turbo"],
+  });
+
+  const calls: string[] = [];
+  const response = await handleComboChat({
+    body: reqBodyTextArray,
+    combo,
+    allCombos: [combo],
+    isModelAvailable: undefined,
+    relayOptions: undefined,
+    signal: undefined,
+    settings: {},
+    log: makeLog(),
+    handleSingleModel: async (_body: unknown, modelStr: string) => {
+      calls.push(modelStr);
+      if (modelStr === "openai/gpt-4") {
+        return Response.json(
+          { error: { message: "requested model is not supported" } },
+          { status: 400 }
+        );
+      }
+      return okResponse(modelStr);
+    },
+  });
+
+  assert.equal(response.status, 200, "combo should advance to second model and return 200");
+  assert.equal(calls.length, 2, "combo should have tried both models");
+  assert.equal(calls[0], "openai/gpt-4", "first model should be tried first");
+  assert.equal(calls[1], "openai/gpt-3.5-turbo", "second model should be tried after 400");
+});