fix(ratelimit): 图片模型被 Codex 文本端点拒绝时不再写模型冷却

gpt-image-* 被误发到 /v1/chat/completions 或 /v1/responses 时，Codex 会稳定
返回 400 codex_plan_gated_model。HandleUpstreamModelNotFound 把它当成账号能力
缺失，给每个账号写 30 分钟 model_rate_limits 冷却。failover 走完整个号池后，
池内每个账号都冷却了该图片模型，于是后续走对端点的 /v1/images/generations
请求也一并 503 no available accounts。

在 codex plan-gated 分支里对图片模型提前返回：仍然 failover 掉本次尝试，
但跳过 SetModelRateLimit 写入，不把号池对正确的生图请求整体下线。

Fixes #4828

backend/internal/service/ratelimit_service.go

@@ -2047,6 +2047,14 @@ func (s *RateLimitService) HandleUpstreamModelNotFound(ctx context.Context, acco
 	case isUpstreamModelNotFoundError(statusCode, responseBody):
 		cooldown, reason = upstreamModelNotFoundCooldown, upstreamModelNotFoundReason
 	case isOpenAIOAuthAccount(account) && isOpenAICodexPlanGatedModelError(statusCode, responseBody):
+		// Image models rejected by the Codex text endpoints are a deterministic
+		// endpoint mismatch, not an account capability gap: the same account still
+		// serves them over /v1/images/*. Cooling down the model here would take the
+		// whole pool offline for correct image requests, so fail the attempt over
+		// without recording a per-model cooldown.
+		if IsGPTImageGenerationModel(requestedModel) {
+			return true
+		}
 		cooldown, reason = upstreamCodexPlanGatedModelCooldown, upstreamCodexPlanGatedModelReason
 	default:
 		return false

backend/internal/service/ratelimit_service_model_not_found_test.go

@@ -382,6 +382,49 @@ func TestRateLimitService_HandleUpstreamError_CodexPlanGatedModelIgnoresAPIKeyAc
 	require.Empty(t, repo.modelRateLimitCalls)
 }
 
+func TestRateLimitService_HandleUpstreamError_CodexPlanGatedImageModelSkipsCooldown(t *testing.T) {
+	for _, model := range []string{"gpt-image-1", "gpt-image-1.5", "gpt-image-2"} {
+		t.Run(model, func(t *testing.T) {
+			repo := &modelNotFoundAccountRepoStub{}
+			svc := &RateLimitService{accountRepo: repo}
+			account := openAICodexPlanGatedOAuthAccount()
+
+			handled := svc.HandleUpstreamError(
+				context.Background(),
+				account,
+				http.StatusBadRequest,
+				http.Header{},
+				[]byte(`{"detail":"The '`+model+`' model is not supported when using Codex with a ChatGPT account."}`),
+				model,
+			)
+
+			require.True(t, handled, "attempt should still fail over")
+			require.Empty(t, repo.modelRateLimitCalls,
+				"image models must not be cooled down: the account still serves them over /v1/images/*")
+			require.Zero(t, repo.tempCalls)
+		})
+	}
+}
+
+func TestRateLimitService_HandleUpstreamError_CodexPlanGatedTextModelStillCoolsDown(t *testing.T) {
+	repo := &modelNotFoundAccountRepoStub{}
+	svc := &RateLimitService{accountRepo: repo}
+	account := openAICodexPlanGatedOAuthAccount()
+
+	handled := svc.HandleUpstreamError(
+		context.Background(),
+		account,
+		http.StatusBadRequest,
+		http.Header{},
+		[]byte(`{"detail":"The 'gpt-5.6-sol' model is not supported when using Codex with a ChatGPT account."}`),
+		"gpt-5.6-sol",
+	)
+
+	require.True(t, handled)
+	require.Len(t, repo.modelRateLimitCalls, 1, "non-image plan-gated models keep the existing cooldown")
+	require.Equal(t, upstreamCodexPlanGatedModelReason, repo.modelRateLimitCalls[0].reason)
+}
+
 func openAICodexPlanGatedOAuthAccount() *Account {
 	return &Account{
 		ID:          202,