🐛 fix: fix model list issue in client mode (lobehub#6240)

* fix model list issue in client mode

* fix tests

src/app/[variants]/(main)/settings/provider/(detail)/[id]/page.tsx

@@ -24,7 +24,7 @@ const Page = async (props: PagePropsWithId) => {
   if (isServerMode) {
     const { userId } = await getUserAuth();
 
-    const { aiProvider } = getServerGlobalConfig();
+    const { aiProvider } = await getServerGlobalConfig();
     const aiInfraRepos = new AiInfraRepos(
       serverDB,
       userId!,

src/config/aiModels/azure.ts

@@ -4,31 +4,84 @@ const azureChatModels: AIChatModelCard[] = [
   {
     abilities: {
       functionCall: true,
+      reasoning: true,
     },
     config: {
-      deploymentName: 'gpt-35-turbo',
+      deploymentName: 'o3-mini',
     },
-    contextWindowTokens: 16_385,
+    contextWindowTokens: 200_000,
     description:
-      'GPT 3.5 Turbo，OpenAI提供的高效模型，适用于聊天和文本生成任务，支持并行函数调用。',
-    displayName: 'GPT 3.5 Turbo',
+      'o3-mini 是我们最新的小型推理模型，在与 o1-mini 相同的成本和延迟目标下提供高智能。',
+    displayName: 'OpenAI o3-mini',
+    id: 'o3-mini',
+    maxOutput: 100_000,
+    pricing: {
+      input: 1.1,
+      output: 4.4,
+    },
+    releasedAt: '2025-01-31',
+    type: 'chat',
+  },
+  {
+    abilities: {
+      reasoning: true,
+    },
+    config: {
+      deploymentName: 'o1-mini',
+    },
+    contextWindowTokens: 128_000,
+    description:
+      'o1-mini是一款针对编程、数学和科学应用场景而设计的快速、经济高效的推理模型。该模型具有128K上下文和2023年10月的知识截止日期。',
+    displayName: 'OpenAI o1-mini',
     enabled: true,
-    id: 'gpt-3.5-turbo',
-    maxOutput: 4096,
+    id: 'o1-mini',
+    maxOutput: 65_536,
+    pricing: {
+      input: 1.1,
+      output: 4.4,
+    },
+    releasedAt: '2024-09-12',
     type: 'chat',
   },
   {
     abilities: {
-      functionCall: true,
+      reasoning: true,
     },
     config: {
-      deploymentName: 'gpt-35-turbo-16k',
+      deploymentName: 'o1',
     },
-    contextWindowTokens: 16_384,
-    description: 'GPT 3.5 Turbo 16k，高容量文本生成模型，适合复杂任务。',
-    displayName: 'GPT 3.5 Turbo',
-    id: 'gpt-3.5-turbo-16k',
-    maxOutput: 4096,
+    contextWindowTokens: 200_000,
+    description:
+      'o1是OpenAI新的推理模型，支持图文输入并输出文本，适用于需要广泛通用知识的复杂任务。该模型具有200K上下文和2023年10月的知识截止日期。',
+    displayName: 'OpenAI o1',
+    enabled: true,
+    id: 'o1',
+    maxOutput: 100_000,
+    pricing: {
+      input: 15,
+      output: 60,
+    },
+    releasedAt: '2024-12-17',
+    type: 'chat',
+  },
+  {
+    abilities: {
+      reasoning: true,
+    },
+    config: {
+      deploymentName: 'o1-preview',
+    },
+    contextWindowTokens: 128_000,
+    description:
+      'o1是OpenAI新的推理模型，适用于需要广泛通用知识的复杂任务。该模型具有128K上下文和2023年10月的知识截止日期。',
+    displayName: 'OpenAI o1-preview',
+    id: 'o1-preview',
+    maxOutput: 32_768,
+    pricing: {
+      input: 15,
+      output: 60,
+    },
+    releasedAt: '2024-09-12',
     type: 'chat',
   },
   {
@@ -37,29 +90,34 @@ const azureChatModels: AIChatModelCard[] = [
       vision: true,
     },
     config: {
-      deploymentName: 'gpt-4-turbo',
+      deploymentName: 'gpt-4o',
     },
     contextWindowTokens: 128_000,
-    description: 'GPT 4 Turbo，多模态模型，提供杰出的语言理解和生成能力，同时支持图像输入。',
-    displayName: 'GPT 4 Turbo',
+    description:
+      'ChatGPT-4o 是一款动态模型，实时更新以保持当前最新版本。它结合了强大的语言理解与生成能力，适合于大规模应用场景，包括客户服务、教育和技术支持。',
+    displayName: 'GPT-4o',
     enabled: true,
-    id: 'gpt-4',
-    maxOutput: 4096,
+    id: 'gpt-4o',
+    pricing: {
+      input: 2.5,
+      output: 10,
+    },
+    releasedAt: '2024-05-13',
     type: 'chat',
   },
+
   {
     abilities: {
       functionCall: true,
       vision: true,
     },
     config: {
-      deploymentName: 'gpt-4o-mini',
+      deploymentName: 'gpt-4-turbo',
     },
     contextWindowTokens: 128_000,
-    description: 'GPT-4o Mini，小型高效模型，具备与GPT-4o相似的卓越性能。',
-    displayName: 'GPT 4o Mini',
-    enabled: true,
-    id: 'gpt-4o-mini',
+    description: 'GPT 4 Turbo，多模态模型，提供杰出的语言理解和生成能力，同时支持图像输入。',
+    displayName: 'GPT 4 Turbo',
+    id: 'gpt-4',
     maxOutput: 4096,
     type: 'chat',
   },
@@ -69,13 +127,13 @@ const azureChatModels: AIChatModelCard[] = [
       vision: true,
     },
     config: {
-      deploymentName: 'gpt-4o',
+      deploymentName: 'gpt-4o-mini',
     },
     contextWindowTokens: 128_000,
-    description: 'GPT-4o 是最新的多模态模型，结合高级文本和图像处理能力。',
-    displayName: 'GPT 4o',
+    description: 'GPT-4o Mini，小型高效模型，具备与GPT-4o相似的卓越性能。',
+    displayName: 'GPT 4o Mini',
     enabled: true,
-    id: 'gpt-4o',
+    id: 'gpt-4o-mini',
     maxOutput: 4096,
     type: 'chat',
   },

src/database/repositories/aiInfra/index.ts

@@ -22,7 +22,7 @@ export class AiInfraRepos {
   private userId: string;
   private db: LobeChatDatabase;
   aiProviderModel: AiProviderModel;
-  private providerConfigs: Record<string, ProviderConfig>;
+  private readonly providerConfigs: Record<string, ProviderConfig>;
   aiModelModel: AiModelModel;
 
   constructor(
@@ -66,6 +66,9 @@ export class AiInfraRepos {
     });
   };
 
+  /**
+   * used in the chat page. to show the enabled providers
+   */
   getUserEnabledProviderList = async () => {
     const list = await this.getAiProviderList();
     return list
@@ -81,6 +84,9 @@ export class AiInfraRepos {
       );
   };
 
+  /**
+   * used in the chat page. to show the enabled models
+   */
   getEnabledModels = async () => {
     const providers = await this.getAiProviderList();
     const enabledProviders = providers.filter((item) => item.enabled);
@@ -129,15 +135,6 @@ export class AiInfraRepos {
     ) as EnabledAiModel[];
   };
 
-  getAiProviderModelList = async (providerId: string) => {
-    const aiModels = await this.aiModelModel.getModelListByProviderId(providerId);
-
-    const defaultModels: AiProviderModelListItem[] =
-      (await this.fetchBuiltinModels(providerId)) || [];
-
-    return mergeArrayById(defaultModels, aiModels) as AiProviderModelListItem[];
-  };
-
   getAiProviderRuntimeState = async (
     decryptor?: DecryptUserKeyVaults,
   ): Promise<AiProviderRuntimeState> => {
@@ -156,6 +153,18 @@ export class AiInfraRepos {
     return { enabledAiModels, enabledAiProviders, runtimeConfig };
   };
 
+  getAiProviderModelList = async (providerId: string) => {
+    const aiModels = await this.aiModelModel.getModelListByProviderId(providerId);
+
+    const defaultModels: AiProviderModelListItem[] =
+      (await this.fetchBuiltinModels(providerId)) || [];
+
+    return mergeArrayById(defaultModels, aiModels) as AiProviderModelListItem[];
+  };
+
+  /**
+   * use in the `/settings/provider/[id]` page
+   */
   getAiProviderDetail = async (id: string, decryptor?: DecryptUserKeyVaults) => {
     const config = await this.aiProviderModel.getAiProviderById(id, decryptor);
 
@@ -171,6 +180,7 @@ export class AiInfraRepos {
     try {
       const { default: providerModels } = await import(`@/config/aiModels/${providerId}`);
 
+      // use the serverModelLists as the defined server model list
       const presetList = this.providerConfigs[providerId]?.serverModelLists || providerModels;
       return (presetList as AIChatModelCard[]).map<AiProviderModelListItem>((m) => ({
         ...m,

src/libs/agent-runtime/openai/__snapshots__/index.test.ts.snap

@@ -22,8 +22,8 @@ exports[`LobeOpenAI > models > should get models 1`] = `
   },
   {
     "contextWindowTokens": 16385,
-    "displayName": "GPT 3.5 Turbo",
-    "enabled": true,
+    "displayName": "GPT-3.5 Turbo",
+    "enabled": false,
     "functionCall": true,
     "id": "gpt-3.5-turbo",
     "reasoning": false,
@@ -39,8 +39,8 @@ exports[`LobeOpenAI > models > should get models 1`] = `
     "vision": false,
   },
   {
-    "contextWindowTokens": 16384,
-    "displayName": "GPT 3.5 Turbo",
+    "contextWindowTokens": undefined,
+    "displayName": undefined,
     "enabled": false,
     "functionCall": true,
     "id": "gpt-3.5-turbo-16k",
@@ -221,7 +221,7 @@ exports[`LobeOpenAI > models > should get models 1`] = `
   {
     "contextWindowTokens": 128000,
     "displayName": "GPT 4 Turbo",
-    "enabled": true,
+    "enabled": false,
     "functionCall": true,
     "id": "gpt-4",
     "reasoning": false,

src/server/globalConfig/index.ts

@@ -12,7 +12,7 @@ import { genServerAiProvidersConfig } from './genServerAiProviderConfig';
 import { parseAgentConfig } from './parseDefaultAgent';
 import { parseFilesConfig } from './parseFilesConfig';
 
-export const getServerGlobalConfig = () => {
+export const getServerGlobalConfig = async () => {
   const { ACCESS_CODES, DEFAULT_AGENT_CONFIG } = getAppConfig();
 
   const config: GlobalServerConfig = {
@@ -38,6 +38,9 @@ export const getServerGlobalConfig = () => {
       ollama: {
         fetchOnClient: !process.env.OLLAMA_PROXY_URL,
       },
+      volcengine: {
+        withDeploymentName: true,
+      },
     }),
     defaultAgent: {
       config: parseAgentConfig(DEFAULT_AGENT_CONFIG),

src/server/routers/lambda/aiModel.ts

@@ -19,7 +19,7 @@ const aiModelProcedure = authedProcedure.use(async (opts) => {
   const { ctx } = opts;
 
   const gateKeeper = await KeyVaultsGateKeeper.initWithEnvKey();
-  const { aiProvider } = getServerGlobalConfig();
+  const { aiProvider } = await getServerGlobalConfig();
 
   return opts.next({
     ctx: {

src/server/routers/lambda/aiProvider.ts

@@ -19,7 +19,7 @@ import { ProviderConfig } from '@/types/user/settings';
 const aiProviderProcedure = authedProcedure.use(async (opts) => {
   const { ctx } = opts;
 
-  const { aiProvider } = getServerGlobalConfig();
+  const { aiProvider } = await getServerGlobalConfig();
 
   const gateKeeper = await KeyVaultsGateKeeper.initWithEnvKey();
   return opts.next({

src/services/aiModel/client.ts

@@ -10,7 +10,13 @@ export class ClientService extends BaseClientService implements IAiModelService
     return new AiModelModel(clientDB as any, this.userId);
   }
   private get aiInfraRepos(): AiInfraRepos {
-    return new AiInfraRepos(clientDB as any, this.userId, {});
+    let config = {};
+
+    if (typeof window !== 'undefined') {
+      config = window.global_serverConfigStore.getState().serverConfig.aiProvider || {};
+    }
+
+    return new AiInfraRepos(clientDB as any, this.userId, config);
   }
 
   createAiModel: IAiModelService['createAiModel'] = async (params) => {