🐛 fix: Custom provider fails when client requests are enabled (#9534)

* ✨ fix: (启用客户端请求，自定义服务商未遵循指定请求格式) 更新 initializeWithClientStore 函数，支持通过选项对象传递 provider 和 payload，增强代码可读性

* ✨ feat: 添加 runtimeProvider 支持，优化模型服务的提供者初始化逻辑

* add test

packages/model-runtime/src/core/ModelRuntime.ts

@@ -41,7 +41,7 @@ export class ModelRuntime {
    *
    * @example - Use without trace
    * ```ts
-   * const agentRuntime = await initializeWithClientStore(provider, payload);
+   * const agentRuntime = await initializeWithClientStore({ provider, payload });
    * const data = payload as ChatStreamPayload;
    * return await agentRuntime.chat(data);
    * ```

packages/types/src/auth.ts

@@ -16,6 +16,8 @@ export interface ClientSecretPayload {
    */
   baseURL?: string;
 
+  runtimeProvider?: string;
+
   azureApiVersion?: string;
 
   awsAccessKeyId?: string;

src/server/modules/ModelRuntime/index.test.ts

@@ -87,6 +87,19 @@ describe('initModelRuntimeWithUserPayload method', () => {
       expect(runtime['_runtime'].baseURL).toBe(jwtPayload.baseURL);
     });
 
+    it('Custom provider should use runtimeProvider to init', async () => {
+      const jwtPayload: ClientSecretPayload = {
+        apiKey: 'user-azure-key',
+        azureApiVersion: '2024-06-01',
+        baseURL: 'user-azure-endpoint',
+        runtimeProvider: ModelProvider.Azure,
+      };
+      const runtime = await initModelRuntimeWithUserPayload('custom-provider', jwtPayload);
+      expect(runtime).toBeInstanceOf(ModelRuntime);
+      expect(runtime['_runtime']).toBeInstanceOf(LobeAzureOpenAI);
+      expect(runtime['_runtime'].baseURL).toBe(jwtPayload.baseURL);
+    });
+
     it('ZhiPu AI provider: with apikey', async () => {
       const jwtPayload: ClientSecretPayload = { apiKey: 'zhipu.user-key' };
       const runtime = await initModelRuntimeWithUserPayload(ModelProvider.ZhiPu, jwtPayload);

src/server/modules/ModelRuntime/index.ts

@@ -128,8 +128,10 @@ export const initModelRuntimeWithUserPayload = (
   payload: ClientSecretPayload,
   params: any = {},
 ) => {
-  return ModelRuntime.initializeWithProvider(provider, {
-    ...getParamsFromPayload(provider, payload),
+  const runtimeProvider = payload.runtimeProvider ?? provider;
+
+  return ModelRuntime.initializeWithProvider(runtimeProvider, {
+    ...getParamsFromPayload(runtimeProvider, payload),
     ...params,
   });
 };

src/services/__tests__/models.test.ts

@@ -1,21 +1,111 @@
-import { Mock, describe, expect, it, vi } from 'vitest';
+import { Mock, beforeEach, describe, expect, it, vi } from 'vitest';
 
+import { aiProviderSelectors } from '@/store/aiInfra';
+
+import { createHeaderWithAuth } from '../_auth';
+import { initializeWithClientStore } from '../chat/clientModelRuntime';
+import { resolveRuntimeProvider } from '../chat/helper';
 import { ModelsService } from '../models';
 
 vi.stubGlobal('fetch', vi.fn());
 
-// 创建一个测试用的 ModelsService 实例
+vi.mock('@/const/version', () => ({
+  isDeprecatedEdition: false,
+}));
 
+vi.mock('../_auth', () => ({
+  createHeaderWithAuth: vi.fn(async () => ({})),
+}));
+
+vi.mock('../chat/helper', () => ({
+  resolveRuntimeProvider: vi.fn((provider: string) => provider),
+}));
+
+vi.mock('../chat/clientModelRuntime', () => ({
+  initializeWithClientStore: vi.fn(),
+}));
+
+vi.mock('@/store/aiInfra', () => ({
+  aiProviderSelectors: {
+    isProviderFetchOnClient: () => () => false,
+  },
+  getAiInfraStoreState: () => ({}),
+}));
+
+vi.mock('@/store/user', () => ({
+  useUserStore: {
+    getState: vi.fn(),
+  },
+}));
+
+vi.mock('@/store/user/selectors', () => ({
+  modelConfigSelectors: {
+    isProviderFetchOnClient: () => () => false,
+  },
+}));
+
+// 创建一个测试用的 ModelsService 实例
 const modelsService = new ModelsService();
 
+const mockedCreateHeaderWithAuth = vi.mocked(createHeaderWithAuth);
+const mockedResolveRuntimeProvider = vi.mocked(resolveRuntimeProvider);
+const mockedInitializeWithClientStore = vi.mocked(initializeWithClientStore);
+
 describe('ModelsService', () => {
+  beforeEach(() => {
+    (fetch as Mock).mockClear();
+    mockedCreateHeaderWithAuth.mockClear();
+    mockedResolveRuntimeProvider.mockReset();
+    mockedResolveRuntimeProvider.mockImplementation((provider: string) => provider);
+    mockedInitializeWithClientStore.mockClear();
+  });
+
   describe('getModels', () => {
-    it('should call the appropriate endpoint for a generic provider', async () => {
-      (fetch as Mock).mockResolvedValueOnce(new Response(JSON.stringify({ models: [] })));
+    it('should call the endpoint for runtime provider when server fetching', async () => {
+      (fetch as Mock).mockResolvedValueOnce(
+        new Response(JSON.stringify({ models: [] }), { status: 200 }),
+      );
 
       await modelsService.getModels('openai');
 
-      expect(fetch).toHaveBeenCalled();
+      expect(mockedResolveRuntimeProvider).toHaveBeenCalledWith('openai');
+      expect(fetch).toHaveBeenCalledWith('/webapi/models/openai', { headers: {} });
+      expect(mockedInitializeWithClientStore).not.toHaveBeenCalled();
+    });
+
+    it('should map custom provider to runtime provider endpoint', async () => {
+      mockedResolveRuntimeProvider.mockImplementation(() => 'openai');
+      (fetch as Mock).mockResolvedValueOnce(
+        new Response(JSON.stringify({ models: [] }), { status: 200 }),
+      );
+
+      await modelsService.getModels('custom-provider');
+
+      expect(mockedResolveRuntimeProvider).toHaveBeenCalledWith('custom-provider');
+      expect(fetch).toHaveBeenCalledWith('/webapi/models/openai', { headers: {} });
+      expect(mockedInitializeWithClientStore).not.toHaveBeenCalled();
+    });
+
+    it('should fetch models on client when isProviderFetchOnClient is true', async () => {
+      // Mock isProviderFetchOnClient to return true
+      const spyIsClient = vi
+        .spyOn(aiProviderSelectors, 'isProviderFetchOnClient')
+        .mockReturnValue(() => true);
+      // Mock initializeWithClientStore to return a runtime with a models() method
+      const mockModels = vi.fn().mockResolvedValue({ models: ['model1', 'model2'] });
+      mockedInitializeWithClientStore.mockResolvedValue({ models: mockModels } as any);
+
+      const result = await modelsService.getModels('openai');
+
+      expect(spyIsClient).toHaveBeenCalledWith('openai');
+      expect(mockedInitializeWithClientStore).toHaveBeenCalledWith({
+        provider: 'openai',
+        runtimeProvider: 'openai',
+      });
+      expect(mockModels).toHaveBeenCalled();
+      expect(result).toEqual({ models: ['model1', 'model2'] });
+
+      spyIsClient.mockRestore();
     });
   });
 });

src/services/_auth.ts

@@ -14,6 +14,8 @@ import { useUserStore } from '@/store/user';
 import { keyVaultsConfigSelectors, userProfileSelectors } from '@/store/user/selectors';
 import { obfuscatePayloadWithXOR } from '@/utils/client/xor-obfuscation';
 
+import { resolveRuntimeProvider } from './chat/helper';
+
 export const getProviderAuthPayload = (
   provider: string,
   keyVaults: OpenAICompatibleKeyVault &
@@ -104,7 +106,12 @@ export const createPayloadWithKeyVaults = (provider: string) => {
     keyVaults = aiProviderSelectors.providerKeyVaults(provider)(useAiInfraStore.getState()) || {};
   }
 
-  return getProviderAuthPayload(provider, keyVaults);
+  const runtimeProvider = resolveRuntimeProvider(provider);
+
+  return {
+    ...getProviderAuthPayload(runtimeProvider, keyVaults as any),
+    runtimeProvider,
+  };
 };
 
 export const createXorKeyVaultsPayload = (provider: string) => {

src/services/chat/clientModelRuntime.test.ts

@@ -108,7 +108,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.OpenAI, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.OpenAI,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeOpenAI);
         expect(runtime['_runtime'].baseURL).toBe('user-openai-endpoint');
@@ -127,7 +130,10 @@ describe('ModelRuntimeOnClient', () => {
           },
         } as UserSettingsState) as unknown as UserStore;
 
-        const runtime = await initializeWithClientStore(ModelProvider.Azure, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Azure,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeAzureOpenAI);
       });
@@ -142,7 +148,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Google, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Google,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeGoogleAI);
       });
@@ -157,7 +166,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Moonshot, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Moonshot,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeMoonshotAI);
       });
@@ -174,7 +186,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Bedrock, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Bedrock,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeBedrockAI);
       });
@@ -189,7 +204,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Ollama, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Ollama,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeOllamaAI);
       });
@@ -204,7 +222,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Perplexity, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Perplexity,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobePerplexityAI);
       });
@@ -219,7 +240,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Anthropic, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Anthropic,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeAnthropicAI);
       });
@@ -234,7 +258,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Mistral, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Mistral,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeMistralAI);
       });
@@ -249,7 +276,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.OpenRouter, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.OpenRouter,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeOpenRouterAI);
       });
@@ -264,7 +294,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.TogetherAI, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.TogetherAI,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeTogetherAI);
       });
@@ -279,7 +312,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.ZeroOne, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.ZeroOne,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeZeroOneAI);
       });
@@ -295,7 +331,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Groq, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Groq,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         const lobeOpenAICompatibleInstance = runtime['_runtime'] as LobeOpenAICompatibleRuntime;
         expect(lobeOpenAICompatibleInstance).toBeInstanceOf(LobeGroq);
@@ -314,7 +353,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.DeepSeek, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.DeepSeek,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeDeepSeekAI);
       });
@@ -329,7 +371,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.Qwen, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.Qwen,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeQwenAI);
       });
@@ -349,7 +394,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as any as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore('unknown' as ModelProvider, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: 'unknown' as ModelProvider,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeOpenAI);
       });
@@ -376,7 +424,10 @@ describe('ModelRuntimeOnClient', () => {
             },
           },
         } as UserSettingsState) as unknown as UserStore;
-        const runtime = await initializeWithClientStore(ModelProvider.ZhiPu, {});
+        const runtime = await initializeWithClientStore({
+          payload: {},
+          provider: ModelProvider.ZhiPu,
+        });
         expect(runtime).toBeInstanceOf(ModelRuntime);
         expect(runtime['_runtime']).toBeInstanceOf(LobeZhipuAI);
       });

src/services/chat/clientModelRuntime.ts

@@ -2,15 +2,26 @@ import { ModelRuntime } from '@lobechat/model-runtime';
 
 import { createPayloadWithKeyVaults } from '../_auth';
 
+export interface InitializeWithClientStoreOptions {
+  payload?: any;
+  provider: string;
+  runtimeProvider?: string;
+}
+
 /**
  * Initializes the AgentRuntime with the client store.
- * @param provider - The provider name.
- * @param payload - Init options
+ * @param options.provider - Provider identifier used to resolve key vaults.
+ * @param options.runtimeProvider - Actual runtime implementation key (defaults to provider).
+ * @param options.payload - Additional initialization payload.
  * @returns The initialized AgentRuntime instance
  *
  * **Note**: if you try to fetch directly, use `fetchOnClient` instead.
  */
-export const initializeWithClientStore = (provider: string, payload?: any) => {
+export const initializeWithClientStore = ({
+  provider,
+  runtimeProvider,
+  payload,
+}: InitializeWithClientStoreOptions) => {
   /**
    * Since #5267, we map parameters for client-fetch in function `getProviderAuthPayload`
    * which called by `createPayloadWithKeyVaults` below.
@@ -26,7 +37,7 @@ export const initializeWithClientStore = (provider: string, payload?: any) => {
    * Configuration override order:
    * payload -> providerAuthPayload -> commonOptions
    */
-  return ModelRuntime.initializeWithProvider(provider, {
+  return ModelRuntime.initializeWithProvider(runtimeProvider ?? provider, {
     ...commonOptions,
     ...providerAuthPayload,
     ...payload,

src/services/chat/helper.ts

@@ -54,3 +54,14 @@ export const isEnableFetchOnClient = (provider: string) => {
     return aiProviderSelectors.isProviderFetchOnClient(provider)(getAiInfraStoreState());
   }
 };
+
+export const resolveRuntimeProvider = (provider: string) => {
+  if (isDeprecatedEdition) return provider;
+
+  const isBuiltin = Object.values(ModelProvider).includes(provider as any);
+  if (isBuiltin) return provider;
+
+  const providerConfig = aiProviderSelectors.providerConfigById(provider)(getAiInfraStoreState());
+
+  return providerConfig?.settings.sdkType || 'openai';
+};

src/services/chat/index.ts

@@ -6,7 +6,7 @@ import { ModelProvider } from 'model-bank';
 
 import { enableAuth } from '@/const/auth';
 import { DEFAULT_AGENT_CONFIG } from '@/const/settings';
-import { isDeprecatedEdition, isDesktop } from '@/const/version';
+import { isDesktop } from '@/const/version';
 import { getSearchConfig } from '@/helpers/getSearchConfig';
 import { createChatToolsEngine, createToolsEngine } from '@/helpers/toolEngineering';
 import { getAgentStoreState } from '@/store/agent';
@@ -38,7 +38,7 @@ import { createHeaderWithAuth } from '../_auth';
 import { API_ENDPOINTS } from '../_url';
 import { initializeWithClientStore } from './clientModelRuntime';
 import { contextEngineering } from './contextEngineering';
-import { findDeploymentName, isEnableFetchOnClient } from './helper';
+import { findDeploymentName, isEnableFetchOnClient, resolveRuntimeProvider } from './helper';
 import { FetchOptions } from './types';
 
 interface GetChatCompletionPayload extends Partial<Omit<ChatStreamPayload, 'messages'>> {
@@ -268,6 +268,8 @@ class ChatService {
       { ...res, apiMode, model },
     );
 
+    const sdkType = resolveRuntimeProvider(provider);
+
     /**
      * Use browser agent runtime
      */
@@ -287,7 +289,7 @@ class ChatService {
        */
       fetcher = async () => {
         try {
-          return await this.fetchOnClient({ payload, provider, signal });
+          return await this.fetchOnClient({ payload, provider, runtimeProvider: sdkType, signal });
         } catch (e) {
           const {
             errorType = ChatErrorType.BadRequest,
@@ -314,17 +316,6 @@ class ChatService {
     const { DEFAULT_MODEL_PROVIDER_LIST } = await import('@/config/modelProviders');
     const providerConfig = DEFAULT_MODEL_PROVIDER_LIST.find((item) => item.id === provider);
 
-    let sdkType = provider;
-    const isBuiltin = Object.values(ModelProvider).includes(provider as any);
-
-    // TODO: remove `!isDeprecatedEdition` condition in V2.0
-    if (!isDeprecatedEdition && !isBuiltin) {
-      const providerConfig =
-        aiProviderSelectors.providerConfigById(provider)(getAiInfraStoreState());
-
-      sdkType = providerConfig?.settings.sdkType || 'openai';
-    }
-
     const userPreferTransitionMode =
       userGeneralSettingsSelectors.transitionMode(getUserStoreState());
 
@@ -461,6 +452,7 @@ class ChatService {
   private fetchOnClient = async (params: {
     payload: Partial<ChatStreamPayload>;
     provider: string;
+    runtimeProvider: string;
     signal?: AbortSignal;
   }) => {
     /**
@@ -471,7 +463,11 @@ class ChatService {
       throw AgentRuntimeError.createError(ChatErrorType.InvalidAccessCode);
     }
 
-    const agentRuntime = await initializeWithClientStore(params.provider, params.payload);
+    const agentRuntime = await initializeWithClientStore({
+      payload: params.payload,
+      provider: params.provider,
+      runtimeProvider: params.runtimeProvider,
+    });
     const data = params.payload as ChatStreamPayload;
 
     return agentRuntime.chat(data, { signal: params.signal });

src/services/models.ts

@@ -8,6 +8,7 @@ import { getMessageError } from '@/utils/fetch';
 
 import { API_ENDPOINTS } from './_url';
 import { initializeWithClientStore } from './chat/clientModelRuntime';
+import { resolveRuntimeProvider } from './chat/helper';
 
 const isEnableFetchOnClient = (provider: string) => {
   // TODO: remove this condition in V2.0
@@ -41,17 +42,22 @@ export class ModelsService {
       headers: { 'Content-Type': 'application/json' },
       provider,
     });
+
+    const runtimeProvider = resolveRuntimeProvider(provider);
     try {
       /**
        * Use browser agent runtime
        */
       const enableFetchOnClient = isEnableFetchOnClient(provider);
       if (enableFetchOnClient) {
-        const agentRuntime = await initializeWithClientStore(provider);
+        const agentRuntime = await initializeWithClientStore({
+          provider,
+          runtimeProvider,
+        });
         return agentRuntime.models();
       }
 
-      const res = await fetch(API_ENDPOINTS.models(provider), { headers });
+      const res = await fetch(API_ENDPOINTS.models(runtimeProvider), { headers });
       if (!res.ok) return;
 
       return res.json();
@@ -77,15 +83,19 @@ export class ModelsService {
         provider,
       });
 
+      const runtimeProvider = resolveRuntimeProvider(provider);
       const enableFetchOnClient = isEnableFetchOnClient(provider);
 
       console.log('enableFetchOnClient：', enableFetchOnClient);
       let res: Response;
       if (enableFetchOnClient) {
-        const agentRuntime = await initializeWithClientStore(provider);
+        const agentRuntime = await initializeWithClientStore({
+          provider,
+          runtimeProvider,
+        });
         res = (await agentRuntime.pullModel({ model }, { signal }))!;
       } else {
-        res = await fetch(API_ENDPOINTS.modelPull(provider), {
+        res = await fetch(API_ENDPOINTS.modelPull(runtimeProvider), {
           body: JSON.stringify({ model }),
           headers,
           method: 'POST',