fix(runtime-fallback): trigger fallback on quota/credit exhaustion (fixes #3519)

src/hooks/runtime-fallback/error-classifier.ts

@@ -169,10 +169,9 @@ export function isRetryableError(error: unknown, retryOnErrors: number[]): boole
   }
 
   if (errorType === "quota_exceeded") {
-    // When a provider signals an auto-retry (e.g. "retrying in ~2 weeks"),
-    // we should still trigger fallback to another model rather than STOP.
-    const hasAutoRetrySignal = /retrying\s+in/i.test(message)
-    return hasAutoRetrySignal
+    // Quota exhaustion means the current model/provider cannot serve requests.
+    // Trigger fallback to the next configured model instead of stopping entirely.
+    return true
   }
 
   if (statusCode && retryOnErrors.includes(statusCode)) {

src/hooks/runtime-fallback/quota-error-classifier.regression.test.ts

@@ -3,7 +3,7 @@ import { describe, expect, test } from "bun:test"
 import { classifyErrorType, isRetryableError } from "./error-classifier"
 
 describe("runtime-fallback quota error regressions", () => {
-  test("classifies subscription quota errors as quota_exceeded and stops retry", () => {
+  test("classifies subscription quota errors as quota_exceeded and triggers fallback", () => {
     //#given
     const error = {
       name: "AI_APICallError",
@@ -16,10 +16,11 @@ describe("runtime-fallback quota error regressions", () => {
 
     //#then
     expect(errorType).toBe("quota_exceeded")
-    expect(retryable).toBe(false)
+    // quota exhaustion should trigger fallback to the next model
+    expect(retryable).toBe(true)
   })
 
-  test("treats HTTP 402 payment required as non-retryable", () => {
+  test("treats HTTP 402 payment required as fallback-eligible", () => {
     //#given
     const error = { statusCode: 402, message: "Payment Required" }
 
@@ -27,7 +28,8 @@ describe("runtime-fallback quota error regressions", () => {
     const retryable = isRetryableError(error, [429, 500, 502, 503, 504])
 
     //#then
-    expect(retryable).toBe(false)
+    // payment failure triggers fallback to a different provider/model
+    expect(retryable).toBe(true)
   })
 
   test("keeps HTTP 429 rate limit retryable", () => {
@@ -41,7 +43,7 @@ describe("runtime-fallback quota error regressions", () => {
     expect(retryable).toBe(true)
   })
 
-  test("classifies quota error names as quota_exceeded without retry", () => {
+  test("classifies quota error names as quota_exceeded and triggers fallback", () => {
     //#given
     const error = { name: "QuotaExceededError", message: "Request failed." }
 
@@ -51,6 +53,7 @@ describe("runtime-fallback quota error regressions", () => {
 
     //#then
     expect(errorType).toBe("quota_exceeded")
-    expect(retryable).toBe(false)
+    // quota errors trigger fallback to next configured model
+    expect(retryable).toBe(true)
   })
 })