refactor(ai-gateway): remove retired free Tencent and LongCat models

apps/web/src/app/api/openrouter/[...path]/route.test.ts

@@ -784,16 +784,19 @@ describe('POST /api/openrouter/v1/chat/completions rules-engine actions', () =>
     expect(mockedUpstreamRequest).not.toHaveBeenCalled();
   });
 
-  it('rejects the disabled LongCat free model before upstream', async () => {
-    const { POST } = await import('./route');
-    const response = await POST(makeRequest(makeBody('meituan/longcat-2.0-free')) as never);
+  it.each(['tencent/hy3:free', 'meituan/longcat-2.0-free'])(
+    'rejects the removed free model %s before upstream',
+    async modelId => {
+      const { POST } = await import('./route');
+      const response = await POST(makeRequest(makeBody(modelId)) as never);
 
-    expect(response.status).toBe(404);
-    expect(await response.json()).toMatchObject({
-      error_type: 'unavailable_model',
-    });
-    expect(mockedUpstreamRequest).not.toHaveBeenCalled();
-  });
+      expect(response.status).toBe(404);
+      expect(await response.json()).toMatchObject({
+        error_type: 'unavailable_model',
+      });
+      expect(mockedUpstreamRequest).not.toHaveBeenCalled();
+    }
+  );
 
   it('rate limits rules-engine rate-limit actions before upstream', async () => {
     mockedRedisGet.mockResolvedValue(cachedRulesEngineAction('rate-limit'));

apps/web/src/app/api/openrouter/models/[provider]/[model]/endpoints/route.test.ts

@@ -87,23 +87,26 @@ describe('GET /api/openrouter/models/[provider]/[model]/endpoints', () => {
     });
   });
 
-  test('returns 404 for unavailable models without reading cached metadata', async () => {
-    mockedGetOpenRouterModelsMetadataFromDatabase.mockClear();
-    const modelId = 'openai/gpt-oss-20b:free';
-
-    const response = await GET(request(modelId), {
-      params: Promise.resolve({ provider: 'openai', model: 'gpt-oss-20b:free' }),
-    });
-
-    expect(response.status).toBe(404);
-    expect(response.headers.get('cache-control')).toBe(
-      'public, max-age=0, s-maxage=60, stale-while-revalidate=60'
-    );
-    await expect(response.json()).resolves.toEqual({
-      error: { message: 'Not Found', code: 404 },
-    });
-    expect(mockedGetOpenRouterModelsMetadataFromDatabase).not.toHaveBeenCalled();
-  });
+  test.each(['openai/gpt-oss-20b:free', 'tencent/hy3:free', 'meituan/longcat-2.0-free'])(
+    'returns 404 for unavailable model %s without reading cached metadata',
+    async modelId => {
+      mockedGetOpenRouterModelsMetadataFromDatabase.mockClear();
+      const [provider, model] = modelId.split('/');
+
+      const response = await GET(request(modelId), {
+        params: Promise.resolve({ provider, model }),
+      });
+
+      expect(response.status).toBe(404);
+      expect(response.headers.get('cache-control')).toBe(
+        'public, max-age=0, s-maxage=60, stale-while-revalidate=60'
+      );
+      await expect(response.json()).resolves.toEqual({
+        error: { message: 'Not Found', code: 404 },
+      });
+      expect(mockedGetOpenRouterModelsMetadataFromDatabase).not.toHaveBeenCalled();
+    }
+  );
 
   test('applies custom pricing to every priced endpoint', async () => {
     const model = {

apps/web/src/lib/ai-gateway/models.test.ts

@@ -3,7 +3,6 @@ import {
   autoFreeModels,
   findKiloExclusiveModel,
   getKiloExclusiveInferenceProviderRestriction,
-  isDisabledKiloExclusiveModel,
   isKiloExclusiveRateLimitedModel,
   kiloExclusiveModels,
   preferredModels,
@@ -20,9 +19,7 @@ import {
   claude_opus_4_6_stealth_model,
 } from './providers/anthropic.constants';
 import { gpt_5_6_sol_discounted_model } from './providers/openai-exclusive';
-import { tencent_hy3_free_model } from './providers/tencent';
 import { gemma_4_26b_a4b_it_free_model } from './providers/google';
-import { longcat_2_free_model } from './providers/longcat';
 import { isUnavailableModel } from './unavailable-models';
 import { getRandomNumber } from './getRandomNumber';
 
@@ -32,7 +29,7 @@ describe('rate-limited Kilo-exclusive models', () => {
       gemma_4_26b_a4b_it_free_model,
     ]);
     expect(isKiloExclusiveRateLimitedModel(gemma_4_26b_a4b_it_free_model.public_id)).toBe(true);
-    expect(isKiloExclusiveRateLimitedModel(tencent_hy3_free_model.public_id)).toBe(false);
+    expect(isKiloExclusiveRateLimitedModel('tencent/hy3:free')).toBe(false);
   });
 });
 
@@ -82,26 +79,16 @@ describe('isFreeModel', () => {
       expect(findKiloExclusiveModel('qwen/qwen3.7-plus')).toBeNull();
     });
 
-    test('disables Tencent Hy3 free and excludes it from Auto Free and preferred models', () => {
-      expect(tencent_hy3_free_model.status).toBe('disabled');
-      expect(findKiloExclusiveModel('tencent/hy3:free')).toBeNull();
-      expect(autoFreeModels.map(({ model }) => model)).not.toContain(
-        tencent_hy3_free_model.public_id
-      );
-      expect(preferredModels).not.toContain(tencent_hy3_free_model.public_id);
-    });
-
-    test('disables LongCat 2.0 free and excludes it from Auto Free and preferred models', async () => {
-      expect(kiloExclusiveModels).toContain(longcat_2_free_model);
-      expect(longcat_2_free_model.status).toBe('disabled');
-      expect(isDisabledKiloExclusiveModel(longcat_2_free_model.public_id)).toBe(true);
-      expect(findKiloExclusiveModel(longcat_2_free_model.public_id)).toBeNull();
-      expect(await isFreeModel(longcat_2_free_model.public_id)).toBe(false);
-      expect(autoFreeModels.map(({ model }) => model)).not.toContain(
-        longcat_2_free_model.public_id
-      );
-      expect(preferredModels).not.toContain(longcat_2_free_model.public_id);
-    });
+    test.each(['tencent/hy3:free', 'meituan/longcat-2.0-free'])(
+      'removes %s from exclusive, Auto Free, and preferred models while keeping it unavailable',
+      modelId => {
+        expect(kiloExclusiveModels.some(model => model.public_id === modelId)).toBe(false);
+        expect(findKiloExclusiveModel(modelId)).toBeNull();
+        expect(isUnavailableModel(modelId)).toBe(true);
+        expect(autoFreeModels.map(({ model }) => model)).not.toContain(modelId);
+        expect(preferredModels).not.toContain(modelId);
+      }
+    );
 
     test.each(['minimax/minimax-m3:free', 'minimax/minimax-m2.7:free'])(
       'inherits %s without an exclusive definition or availability restriction',
@@ -392,10 +379,8 @@ describe('getKiloExclusiveInferenceProviderRestriction', () => {
     ).toEqual(new Set(['openai']));
   });
 
-  test('does not treat disabled or unrestricted exclusives or unknown ids as restricted', () => {
-    expect(
-      getKiloExclusiveInferenceProviderRestriction(tencent_hy3_free_model.public_id)
-    ).toBeUndefined();
+  test('does not treat removed or unrestricted exclusives or unknown ids as restricted', () => {
+    expect(getKiloExclusiveInferenceProviderRestriction('tencent/hy3:free')).toBeUndefined();
     expect(
       getKiloExclusiveInferenceProviderRestriction(gemma_4_26b_a4b_it_free_model.public_id)
     ).toBeUndefined();

apps/web/src/lib/ai-gateway/models.ts

@@ -23,8 +23,6 @@ import { KIMI_CURRENT_MODEL_ID } from '@/lib/ai-gateway/providers/moonshotai';
 import { gemma_4_26b_a4b_it_free_model, isGeminiModel } from '@/lib/ai-gateway/providers/google';
 import { qwen36_plus_stealth_model } from '@/lib/ai-gateway/providers/qwen';
 import { stepfun_37_flash_free_model } from '@/lib/ai-gateway/providers/stepfun';
-import { tencent_hy3_free_model } from '@/lib/ai-gateway/providers/tencent';
-import { longcat_2_free_model } from '@/lib/ai-gateway/providers/longcat';
 import { isGrokModel } from '@/lib/ai-gateway/providers/xai';
 import { isClaudeModel } from '@/lib/ai-gateway/providers/anthropic.constants';
 import { GPT_CURRENT_MODEL_ID, isOpenAiModel } from '@/lib/ai-gateway/providers/openai';
@@ -52,29 +50,11 @@ export const autoFreeModels: ReadonlyArray<AutoFreeModel> = [
         } satisfies AutoFreeModel,
       ]
     : []),
-  ...(tencent_hy3_free_model.status === 'public'
-    ? [
-        {
-          model: tencent_hy3_free_model.public_id,
-          weight: 1,
-          reasoning: { enabled: true, effort: 'high' },
-        } satisfies AutoFreeModel,
-      ]
-    : []),
   {
     model: 'poolside/laguna-s-2.1:free',
     weight: 1,
     reasoning: { enabled: true, effort: 'high' },
   } satisfies AutoFreeModel,
-  ...(longcat_2_free_model.status === 'public'
-    ? [
-        {
-          model: longcat_2_free_model.public_id,
-          weight: 1,
-          reasoning: { enabled: true, effort: 'high' },
-        } satisfies AutoFreeModel,
-      ]
-    : []),
   {
     model: 'minimax/minimax-m3:free',
     weight: 1,
@@ -152,8 +132,6 @@ export const kiloExclusiveModels = [
   claude_sonnet_4_6_stealth_model,
   claude_opus_4_6_stealth_model,
   stepfun_37_flash_free_model,
-  tencent_hy3_free_model,
-  longcat_2_free_model,
 ] as KiloExclusiveModel[];
 
 export function isKiloStealthModel(model: string): boolean {

apps/web/src/lib/ai-gateway/providers/longcat.ts

@@ -1,20 +1,3 @@
-import type { KiloExclusiveModel } from '@/lib/ai-gateway/providers/kilo-exclusive-model';
-
 export function isLongCatModel(requestedModel: string) {
   return requestedModel.includes('longcat');
 }
-
-export const longcat_2_free_model: KiloExclusiveModel = {
-  public_id: 'meituan/longcat-2.0-free',
-  display_name: 'Meituan: LongCat 2.0 (free)',
-  description:
-    'LongCat 2.0 is a sparse mixture-of-experts language model from Meituan, with 48B active parameters out of 1.6T total. It is suited for coding, repository-level changes, long-horizon problem solving, and agentic workflows. Available free in Kilo for a limited time.',
-  context_length: 1_048_756,
-  max_completion_tokens: 131_072,
-  status: 'disabled',
-  flags: ['reasoning'],
-  gateway: 'longcat',
-  internal_id: 'LongCat-2.0',
-  pricing: null,
-  inference_provider_restriction: [],
-};

apps/web/src/lib/ai-gateway/unavailable-models.test.ts

@@ -12,6 +12,7 @@ describe('unavailable models', () => {
     expect(isUnavailableModel('openai/gpt-oss-20b')).toBe(false);
     expect(isUnavailableModel('tencent/hy3:free')).toBe(true);
     expect(isUnavailableModel('tencent/hy3')).toBe(false);
+    expect(isUnavailableModel('meituan/longcat-2.0-free')).toBe(true);
   });
 
   test('matches normalized families for provider metadata', () => {

apps/web/src/lib/ai-gateway/unavailable-models.ts

@@ -3,6 +3,7 @@ import { normalizeModelId } from '@/lib/ai-gateway/model-utils';
 const unavailableModelIds: ReadonlySet<string> = new Set([
   'google/gemma-4-26b-a4b-it:free', // usable through kilo-auto
   'google/gemma-4-31b-it:free',
+  'meituan/longcat-2.0-free',
   'nvidia/nemotron-3-nano-30b-a3b:free',
   'nvidia/nemotron-nano-12b-v2-vl:free',
   'nvidia/nemotron-nano-9b-v2:free',

apps/web/src/tests/openrouter-models-config.test.ts

@@ -7,7 +7,6 @@ import {
 import { GPT_CURRENT_MODEL_ID } from '@/lib/ai-gateway/providers/openai';
 import { gpt_5_6_sol_discounted_model } from '@/lib/ai-gateway/providers/openai-exclusive';
 import { QWEN37_PLUS_MODEL_ID } from '@/lib/ai-gateway/providers/qwen';
-import { tencent_hy3_free_model } from '@/lib/ai-gateway/providers/tencent';
 
 describe('OpenRouter Models Config', () => {
   test('preferred models should contain expected models', () => {
@@ -28,7 +27,8 @@ describe('OpenRouter Models Config', () => {
       'stealth/qwen3.6-plus',
       QWEN37_PLUS_MODEL_ID,
       'deepseek/deepseek-v4-pro',
-      tencent_hy3_free_model.public_id,
+      'tencent/hy3:free',
+      'meituan/longcat-2.0-free',
     ];
 
     supersededModels.forEach(model => {

apps/web/src/tests/openrouter-models.test.ts

@@ -282,6 +282,36 @@ describe('GET /api/openrouter/models', () => {
     expect(Array.isArray(responseData.data)).toBe(true);
   });
 
+  test('excludes retired free Tencent and LongCat models advertised upstream while retaining paid Tencent', async () => {
+    const original = mockOpenRouterModels.data.find(model => model.id === 'some-other-model');
+    if (!original) throw new Error('Expected catalog fixture');
+    const upstream = {
+      data: [
+        ...mockOpenRouterModels.data,
+        { ...original, id: 'tencent/hy3', name: 'Tencent: HY3' },
+        ...['tencent/hy3:free', 'meituan/longcat-2.0-free'].map(id => ({
+          ...original,
+          id,
+          name: id,
+          pricing: { ...original.pricing, prompt: '0', completion: '0' },
+        })),
+      ],
+    };
+    global.fetch = jest
+      .fn<ReturnType<typeof fetch>, Parameters<typeof fetch>>()
+      .mockResolvedValue(createMockResponse({ jsonData: upstream }));
+
+    const response = await GET(createTestRequest('/api/openrouter/models'));
+    const responseData = OpenRouterModelsResponseSchema.parse(await response.json());
+    const modelIds = responseData.data.map(model => model.id);
+
+    expect(captureException).not.toHaveBeenCalled();
+    expect(response.status).toBe(200);
+    expect(modelIds).not.toContain('tencent/hy3:free');
+    expect(modelIds).not.toContain('meituan/longcat-2.0-free');
+    expect(modelIds).toContain('tencent/hy3');
+  });
+
   test('should include publishable Terminal Bench summaries for canonical models', async () => {
     const request = createTestRequest('/api/openrouter/models');