@lobehub/chat 0.155.9 → 0.156.1

package/src/app/(main)/settings/llm/Anthropic/index.tsx

@@ -14,7 +14,9 @@ const AnthropicProvider = memo(() => {
     <ProviderConfig
       checkModel={'claude-3-haiku-20240307'}
       provider={ModelProvider.Anthropic}
-      showEndpoint
+      proxyUrl={{
+        placeholder: 'https://api.anthropic.com',
+      }}
       title={<Anthropic.Text color={isDarkMode ? undefined : Claude.colorPrimary} size={15} />}
     />
   );

package/src/app/(main)/settings/llm/Google/index.tsx

@@ -14,7 +14,9 @@ const GoogleProvider = memo(() => {
     <ProviderConfig
       checkModel={'gemini-pro'}
       provider={ModelProvider.Google}
-      showEndpoint
+      proxyUrl={{
+        placeholder: 'https://generativelanguage.googleapis.com',
+      }}
       title={
         <Flexbox align={'center'} gap={8} horizontal>
           <Google.BrandColor size={22} />

package/src/app/(main)/settings/llm/Groq/index.tsx

@@ -15,6 +15,9 @@ const GroqProvider = memo(() => {
     <ProviderConfig
       checkModel={'gemma-7b-it'}
       provider={ModelProvider.Groq}
+      proxyUrl={{
+        placeholder: 'https://api.groq.com/openai/v1',
+      }}
       title={<Groq.Text color={theme.isDarkMode ? theme.colorText : Groq.colorPrimary} size={20} />}
     />
   );

package/src/app/(main)/settings/llm/Ollama/index.tsx

@@ -10,21 +10,25 @@ import ProviderConfig from '../components/ProviderConfig';
 import Checker from './Checker';
 
 const OllamaProvider = memo(() => {
-  const { t } = useTranslation('setting');
+  const { t } = useTranslation('modelProvider');
 
   return (
     <ProviderConfig
       checkerItem={{
         children: <Checker />,
-        desc: t('llm.checker.ollamaDesc'),
-        label: t('llm.checker.title'),
+        desc: t('ollama.checker.desc'),
+        label: t('ollama.checker.title'),
         minWidth: undefined,
       }}
       modelList={{ showModelFetcher: true }}
       provider={ModelProvider.Ollama}
+      proxyUrl={{
+        desc: t('ollama.endpoint.desc'),
+        placeholder: 'http://127.0.0.1:11434',
+        title: t('ollama.endpoint.title'),
+      }}
       showApiKey={false}
       showBrowserRequest
-      showEndpoint
       title={<Ollama.Combine size={28} />}
     />
   );

package/src/app/(main)/settings/llm/OpenAI/index.tsx

@@ -14,9 +14,13 @@ const OpenAIProvider = memo(() => {
     <ProviderConfig
       modelList={{ showModelFetcher: true }}
       provider={'openai'}
+      proxyUrl={
+        showOpenAIProxyUrl && {
+          placeholder: 'https://api.openai.com/v1',
+        }
+      }
       showApiKey={showOpenAIApiKey}
       showBrowserRequest
-      showEndpoint={showOpenAIProxyUrl}
       title={<OpenAI.Combine size={24} />}
     />
   );

package/src/app/(main)/settings/llm/Perplexity/index.tsx

@@ -12,6 +12,9 @@ const PerplexityProvider = memo(() => {
     <ProviderConfig
       checkModel={'pplx-7b-chat'}
       provider={ModelProvider.Perplexity}
+      proxyUrl={{
+        placeholder: 'https://api.perplexity.ai',
+      }}
       title={<Perplexity.Combine size={24} type={'color'} />}
     />
   );

package/src/app/(main)/settings/llm/components/ProviderConfig/index.tsx

@@ -57,9 +57,15 @@ interface ProviderConfigProps {
     showModelFetcher?: boolean;
   };
   provider: GlobalLLMProviderKey;
+  proxyUrl?:
+    | {
+        desc?: string;
+        placeholder: string;
+        title?: string;
+      }
+    | false;
   showApiKey?: boolean;
   showBrowserRequest?: boolean;
-  showEndpoint?: boolean;
   title: ReactNode;
 }
 
@@ -67,7 +73,7 @@ const ProviderConfig = memo<ProviderConfigProps>(
   ({
     apiKeyItems,
     provider,
-    showEndpoint,
+    proxyUrl,
     showApiKey = true,
     checkModel,
     canDeactivate = true,
@@ -112,14 +118,13 @@ const ProviderConfig = memo<ProviderConfigProps>(
           },
         ];
 
+    const showEndpoint = !!proxyUrl;
     const formItems = [
       ...apiKeyItem,
       showEndpoint && {
-        children: (
-          <Input allowClear placeholder={modelT(`${provider}.endpoint.placeholder` as any)} />
-        ),
-        desc: modelT(`${provider}.endpoint.desc` as any),
-        label: modelT(`${provider}.endpoint.title` as any),
+        children: <Input allowClear placeholder={proxyUrl?.placeholder} />,
+        desc: proxyUrl?.desc || t('llm.proxyUrl.desc'),
+        label: proxyUrl?.title || t('llm.proxyUrl.title'),
         name: [LLMProviderConfigKey, provider, LLMProviderBaseUrlKey],
       },
       (showBrowserRequest || (showEndpoint && isProviderEndpointNotEmpty)) && {

package/src/app/api/chat/agentRuntime.ts

@@ -85,76 +85,71 @@ const getLlmOptionsFromPayload = (provider: string, payload: JWTPayload) => {
         accessKeySecret = payload?.awsSecretAccessKey;
         region = payload?.awsRegion;
       }
-      return {
-        accessKeyId,
-        accessKeySecret,
-        region,
-      };
+      return { accessKeyId, accessKeySecret, region };
     }
     case ModelProvider.Ollama: {
       const { OLLAMA_PROXY_URL } = getServerConfig();
       const baseURL = payload?.endpoint || OLLAMA_PROXY_URL;
-      return {
-        baseURL,
-      };
+      return { baseURL };
     }
     case ModelProvider.Perplexity: {
-      const { PERPLEXITY_API_KEY } = getServerConfig();
+      const { PERPLEXITY_API_KEY, PERPLEXITY_PROXY_URL } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || PERPLEXITY_API_KEY);
-      return {
-        apiKey,
-      };
+      const baseURL = payload?.endpoint || PERPLEXITY_PROXY_URL;
+
+      return { apiKey, baseURL };
     }
     case ModelProvider.Anthropic: {
       const { ANTHROPIC_API_KEY, ANTHROPIC_PROXY_URL } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || ANTHROPIC_API_KEY);
       const baseURL = payload?.endpoint || ANTHROPIC_PROXY_URL;
-      return {
-        apiKey,
-        baseURL,
-      };
+
+      return { apiKey, baseURL };
     }
     case ModelProvider.Minimax: {
       const { MINIMAX_API_KEY } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || MINIMAX_API_KEY);
-      return {
-        apiKey,
-      };
+
+      return { apiKey };
     }
     case ModelProvider.Mistral: {
       const { MISTRAL_API_KEY } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || MISTRAL_API_KEY);
-      return {
-        apiKey,
-      };
+
+      return { apiKey };
     }
     case ModelProvider.Groq: {
-      const { GROQ_API_KEY } = getServerConfig();
+      const { GROQ_API_KEY, GROQ_PROXY_URL } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || GROQ_API_KEY);
-      return {
-        apiKey,
-      };
+      const baseURL = payload?.endpoint || GROQ_PROXY_URL;
+
+      return { apiKey, baseURL };
     }
     case ModelProvider.OpenRouter: {
       const { OPENROUTER_API_KEY } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || OPENROUTER_API_KEY);
-      return {
-        apiKey,
-      };
+
+      return { apiKey };
     }
     case ModelProvider.TogetherAI: {
       const { TOGETHERAI_API_KEY } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || TOGETHERAI_API_KEY);
-      return {
-        apiKey,
-      };
+
+      return { apiKey };
     }
     case ModelProvider.ZeroOne: {
       const { ZEROONE_API_KEY } = getServerConfig();
+
       const apiKey = apiKeyManager.pick(payload?.apiKey || ZEROONE_API_KEY);
-      return {
-        apiKey,
-      };
+
+      return { apiKey };
     }
   }
 };

package/src/config/modelProviders/groq.ts

@@ -3,6 +3,12 @@ import { ModelProviderCard } from '@/types/llm';
 // ref https://console.groq.com/docs/models
 const Groq: ModelProviderCard = {
   chatModels: [
+    {
+      displayName: 'LLaMA3-3-70B',
+      enabled: true,
+      id: 'llama3-70b-8192',
+      tokens: 8192,
+    },
     {
       displayName: 'Mixtral-8x7b-Instruct-v0.1',
       enabled: true,
@@ -21,12 +27,6 @@ const Groq: ModelProviderCard = {
       id: 'llama3-8b-8192',
       tokens: 8192,
     },
-    {
-      displayName: 'LLaMA3-3-70B',
-      enabled: true,
-      id: 'llama3-70b-8192',
-      tokens: 8192,
-    },
     {
       displayName: 'LLaMA2-70b-chat',
       id: 'llama2-70b-4096',

package/src/config/modelProviders/togetherai.ts

@@ -6,90 +6,68 @@ const TogetherAI: ModelProviderCard = {
     {
       displayName: 'Deepseek Coder Instruct (33B)',
       enabled: true,
-      functionCall: false,
       id: 'deepseek-ai/deepseek-coder-33b-instruct',
       tokens: 16_384,
-      vision: false,
     },
     {
       displayName: 'Phind Code LLaMA v2 (34B)',
       enabled: true,
-      functionCall: false,
       id: 'Phind/Phind-CodeLlama-34B-v2',
       tokens: 16_384,
-      vision: false,
     },
     {
       displayName: 'Gemma Instruct (2B)',
       enabled: true,
-      functionCall: false,
       id: 'google/gemma-2b-it',
       tokens: 8192,
-      vision: false,
     },
     {
       displayName: 'LLaMA-2 Chat (13B)',
       enabled: true,
-      functionCall: false,
       id: 'meta-llama/Llama-2-13b-chat-hf',
       tokens: 4096,
-      vision: false,
     },
     {
       displayName: '01-ai Yi Chat (34B)',
       enabled: true,
-      functionCall: false,
       id: 'zero-one-ai/Yi-34B-Chat',
       tokens: 4096,
-      vision: false,
     },
     {
       displayName: 'Mixtral-8x7B Instruct (46.7B)',
       enabled: true,
-      functionCall: false,
       id: 'mistralai/Mixtral-8x7B-Instruct-v0.1',
       tokens: 32_768,
-      vision: false,
     },
     {
       displayName: 'Nous Hermes 2 - Mixtral 8x7B-DPO (46.7B)',
       enabled: true,
-      functionCall: false,
       id: 'NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO',
       tokens: 32_768,
-      vision: false,
     },
     {
       displayName: 'Nous Hermes-2 Yi (34B)',
       enabled: true,
-      functionCall: false,
       id: 'NousResearch/Nous-Hermes-2-Yi-34B',
       tokens: 4096,
-      vision: false,
     },
     {
       displayName: 'Qwen 1.5 Chat (7B)',
       enabled: true,
-      functionCall: false,
       id: 'Qwen/Qwen1.5-7B-Chat',
       tokens: 32_768,
-      vision: false,
     },
     {
       displayName: 'Qwen 1.5 Chat (14B)',
       enabled: true,
-      functionCall: false,
       id: 'Qwen/Qwen1.5-14B-Chat',
       tokens: 32_768,
-      vision: false,
     },
     {
       displayName: 'Qwen 1.5 Chat (72B)',
       enabled: true,
-      functionCall: false,
       id: 'Qwen/Qwen1.5-72B-Chat',
       tokens: 32_768,
-      vision: false,
     },
   ],
   id: 'togetherai',

package/src/config/server/provider.ts

@@ -37,6 +37,7 @@ declare global {
       // Perplexity Provider
       ENABLED_PERPLEXITY?: string;
       PERPLEXITY_API_KEY?: string;
+      PERPLEXITY_PROXY_URL?: string;
 
       // Anthropic Provider
       ENABLED_ANTHROPIC?: string;
@@ -54,6 +55,7 @@ declare global {
       // Groq Provider
       ENABLED_GROQ?: string;
       GROQ_API_KEY?: string;
+      GROQ_PROXY_URL?: string;
 
       // OpenRouter Provider
       ENABLED_OPENROUTER?: string;
@@ -172,6 +174,7 @@ export const getProviderConfig = () => {
 
     ENABLED_PERPLEXITY: !!PERPLEXITY_API_KEY,
     PERPLEXITY_API_KEY,
+    PERPLEXITY_PROXY_URL: process.env.PERPLEXITY_PROXY_URL,
 
     ENABLED_ANTHROPIC: !!ANTHROPIC_API_KEY,
     ANTHROPIC_API_KEY,
@@ -197,6 +200,7 @@ export const getProviderConfig = () => {
     MOONSHOT_PROXY_URL: process.env.MOONSHOT_PROXY_URL,
 
     ENABLED_GROQ: !!GROQ_API_KEY,
+    GROQ_PROXY_URL: process.env.GROQ_PROXY_URL,
     GROQ_API_KEY,
 
     ENABLED_ZEROONE: !!ZEROONE_API_KEY,

package/src/libs/agent-runtime/anthropic/index.ts

@@ -27,21 +27,11 @@ export class LobeAnthropicAI implements LobeRuntimeAI {
   }
 
   async chat(payload: ChatStreamPayload, options?: ChatCompetitionOptions) {
-    const { messages, model, max_tokens, temperature, top_p } = payload;
-    const system_message = messages.find((m) => m.role === 'system');
-    const user_messages = messages.filter((m) => m.role !== 'system');
-
     try {
+      const anthropicPayload = this.buildAnthropicPayload(payload);
+
       const response = await this.client.messages.create(
-        {
-          max_tokens: max_tokens || 4096,
-          messages: buildAnthropicMessages(user_messages),
-          model: model,
-          stream: true,
-          system: system_message?.content as string,
-          temperature: temperature,
-          top_p: top_p,
-        },
+        { ...anthropicPayload, stream: true },
         { signal: options?.signal },
       );
 
@@ -71,6 +61,15 @@ export class LobeAnthropicAI implements LobeRuntimeAI {
               provider: ModelProvider.Anthropic,
             });
           }
+
+          case 403: {
+            throw AgentRuntimeError.chat({
+              endpoint: desensitizedEndpoint,
+              error: error as any,
+              errorType: AgentRuntimeErrorType.LocationNotSupportError,
+              provider: ModelProvider.Anthropic,
+            });
+          }
           default: {
             break;
           }
@@ -84,6 +83,22 @@ export class LobeAnthropicAI implements LobeRuntimeAI {
       });
     }
   }
+
+  private buildAnthropicPayload(payload: ChatStreamPayload) {
+    const { messages, model, max_tokens, temperature, top_p } = payload;
+    const system_message = messages.find((m) => m.role === 'system');
+    const user_messages = messages.filter((m) => m.role !== 'system');
+
+    return {
+      max_tokens: max_tokens || 4096,
+      messages: buildAnthropicMessages(user_messages),
+      model: model,
+      stream: true,
+      system: system_message?.content as string,
+      temperature: temperature,
+      top_p: top_p,
+    };
+  }
 }
 
 export default LobeAnthropicAI;

package/src/libs/agent-runtime/azureOpenai/index.test.ts

@@ -163,4 +163,57 @@ describe('LobeAzureOpenAI', () => {
       });
     });
   });
+
+  describe('private method', () => {
+
+    describe('tocamelCase', () => {
+      it('should convert string to camel case', () => {
+        const key = 'image_url';
+
+        const camelCaseKey = instance['tocamelCase'](key);
+
+        expect(camelCaseKey).toEqual('imageUrl');
+      });
+    });
+
+    describe('camelCaseKeys', () => {
+      it('should convert object keys to camel case', () => {
+        const obj = {
+          "frequency_penalty": 0,
+          "messages": [
+            {
+              "role": "user",
+              "content": [
+                {
+                  "type": "image_url",
+                  "image_url": {
+                    "url": "<image URL>"
+                  }
+                }
+              ]
+            }
+          ]
+        };
+
+        const newObj = instance['camelCaseKeys'](obj);
+
+        expect(newObj).toEqual({
+          "frequencyPenalty": 0,
+          "messages": [
+            {
+              "role": "user",
+              "content": [
+                {
+                  "type": "image_url",
+                  "imageUrl": {
+                    "url": "<image URL>"
+                  }
+                }
+              ]
+            }
+          ]
+        });
+      });
+    });
+  })
 });

package/src/libs/agent-runtime/azureOpenai/index.ts

@@ -28,7 +28,8 @@ export class LobeAzureOpenAI implements LobeRuntimeAI {
 
   async chat(payload: ChatStreamPayload, options?: ChatCompetitionOptions) {
     // ============  1. preprocess messages   ============ //
-    const { messages, model, ...params } = payload;
+    const camelCasePayload = this.camelCaseKeys(payload);
+    const { messages, model, maxTokens = 2048, ...params } = camelCasePayload;
 
     // ============  2. send api   ============ //
 
@@ -36,7 +37,7 @@ export class LobeAzureOpenAI implements LobeRuntimeAI {
       const response = await this.client.streamChatCompletions(
         model,
         messages as ChatRequestMessage[],
-        { ...params, abortSignal: options?.signal } as GetChatCompletionsOptions,
+        { ...params, abortSignal: options?.signal, maxTokens } as GetChatCompletionsOptions,
       );
 
       const stream = OpenAIStream(response as any);
@@ -77,4 +78,30 @@ export class LobeAzureOpenAI implements LobeRuntimeAI {
       });
     }
   }
+
+  // Convert object keys to camel case, copy from `@azure/openai` in `node_modules/@azure/openai/dist/index.cjs`
+  private camelCaseKeys = (obj: any): any => {
+    if (typeof obj !== "object" || !obj)
+      return obj;
+    if (Array.isArray(obj)) {
+      return obj.map((v) => this.camelCaseKeys(v));
+    }
+    else {
+      for (const key of Object.keys(obj)) {
+        const value = obj[key];
+        const newKey = this.tocamelCase(key);
+        if (newKey !== key) {
+          delete obj[key];
+        }
+        obj[newKey] = typeof obj[newKey] === "object" ? this.camelCaseKeys(value) : value;
+      }
+      return obj;
+    }
+  }
+
+  private tocamelCase = (str: string) => {
+    return str
+      .toLowerCase()
+      .replaceAll(/(_[a-z])/g, (group) => group.toUpperCase().replace("_", ""));
+  }
 }

package/src/libs/agent-runtime/groq/index.ts

@@ -4,6 +4,13 @@ import { LobeOpenAICompatibleFactory } from '../utils/openaiCompatibleFactory';
 
 export const LobeGroq = LobeOpenAICompatibleFactory({
   baseURL: 'https://api.groq.com/openai/v1',
+  chatCompletion: {
+    handleError: (error) => {
+      // 403 means the location is not supporteds
+      if (error.status === 403)
+        return { error, errorType: AgentRuntimeErrorType.LocationNotSupportError };
+    },
+  },
   debug: {
     chatCompletion: () => process.env.DEBUG_GROQ_CHAT_COMPLETION === '1',
   },

package/src/libs/agent-runtime/perplexity/index.test.ts

@@ -2,7 +2,7 @@
 import OpenAI from 'openai';
 import { Mock, afterEach, beforeEach, describe, expect, it, vi } from 'vitest';
 
-import { ChatStreamCallbacks } from '@/libs/agent-runtime';
+import { ChatStreamCallbacks, LobeOpenAICompatibleRuntime } from '@/libs/agent-runtime';
 
 import * as debugStreamModule from '../utils/debugStream';
 import { LobePerplexityAI } from './index';
@@ -15,7 +15,7 @@ const invalidErrorType = 'InvalidPerplexityAPIKey';
 // Mock the console.error to avoid polluting test output
 vi.spyOn(console, 'error').mockImplementation(() => {});
 
-let instance: LobePerplexityAI;
+let instance: LobeOpenAICompatibleRuntime;
 
 beforeEach(() => {
   instance = new LobePerplexityAI({ apiKey: 'test' });