@lobehub/chat 1.111.6 → 1.111.8

package/docker-compose/setup.sh

@@ -450,7 +450,7 @@ ENV_EXAMPLES=(
     "$SUB_DIR/.env.example"
 )
 # Default values
-CASDOOR_PASSWORD="123"
+CASDOOR_PASSWORD="pswd123"
 CASDOOR_SECRET="CASDOOR_SECRET"
 MINIO_ROOT_PASSWORD="YOUR_MINIO_PASSWORD"
 CASDOOR_HOST="localhost:8000"
@@ -657,10 +657,10 @@ section_regenerate_secrets() {
     CASDOOR_PASSWORD=$(generate_key 10)
     if [ $? -ne 0 ]; then
         echo $(show_message "security_secrect_regenerate_failed") "CASDOOR_PASSWORD"
-        CASDOOR_PASSWORD="123"
+        CASDOOR_PASSWORD="pswd123"
     else
         # replace `password` in init_data.json
-        sed "${SED_INPLACE_ARGS[@]}" "s/"123"/${CASDOOR_PASSWORD}/" init_data.json
+        sed "${SED_INPLACE_ARGS[@]}" "s/"pswd123"/${CASDOOR_PASSWORD}/" init_data.json
         if [ $? -ne 0 ]; then
             echo $(show_message "security_secrect_regenerate_failed") "CASDOOR_PASSWORD in \`init_data.json\`"
         fi

package/locales/ar/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "شدة الاستدلال"
     },
+    "textVerbosity": {
+      "title": "مستوى تفصيل المخرجات النصية"
+    },
     "thinking": {
       "title": "مفتاح التفكير العميق"
     },

package/locales/bg-BG/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Интензитет на разсъждение"
     },
+    "textVerbosity": {
+      "title": "Ниво на подробност на изходния текст"
+    },
     "thinking": {
       "title": "Превключвател за дълбоко мислене"
     },

package/locales/de-DE/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Argumentationsstärke"
     },
+    "textVerbosity": {
+      "title": "Detaillierungsgrad des Ausgabetextes"
+    },
     "thinking": {
       "title": "Tiefdenk-Schalter"
     },

package/locales/en-US/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Reasoning Intensity"
     },
+    "textVerbosity": {
+      "title": "Output Text Detail Level"
+    },
     "thinking": {
       "title": "Deep Thinking Switch"
     },

package/locales/es-ES/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Intensidad de razonamiento"
     },
+    "textVerbosity": {
+      "title": "Nivel de detalle del texto de salida"
+    },
     "thinking": {
       "title": "Interruptor de pensamiento profundo"
     },

package/locales/fa-IR/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "شدت استدلال"
     },
+    "textVerbosity": {
+      "title": "سطح جزئیات متن خروجی"
+    },
     "thinking": {
       "title": "کلید تفکر عمیق"
     },

package/locales/fr-FR/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Intensité du raisonnement"
     },
+    "textVerbosity": {
+      "title": "Niveau de détail du texte de sortie"
+    },
     "thinking": {
       "title": "Interrupteur de réflexion approfondie"
     },

package/locales/it-IT/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Intensità del ragionamento"
     },
+    "textVerbosity": {
+      "title": "Livello di dettaglio dell'output testuale"
+    },
     "thinking": {
       "title": "Interruttore di pensiero profondo"
     },

package/locales/ja-JP/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "推論の強度"
     },
+    "textVerbosity": {
+      "title": "出力テキストの詳細レベル"
+    },
     "thinking": {
       "title": "深い思考のスイッチ"
     },

package/locales/ko-KR/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "추론 강도"
     },
+    "textVerbosity": {
+      "title": "출력 텍스트 상세 수준"
+    },
     "thinking": {
       "title": "심층 사고 스위치"
     },

package/locales/nl-NL/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Redeneringsinspanning"
     },
+    "textVerbosity": {
+      "title": "Gedetailleerdheid van de uitvoer"
+    },
     "thinking": {
       "title": "Diepdenkschakelaar"
     },

package/locales/pl-PL/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Intensywność rozumowania"
     },
+    "textVerbosity": {
+      "title": "Poziom szczegółowości tekstu wyjściowego"
+    },
     "thinking": {
       "title": "Przełącznik głębokiego myślenia"
     },

package/locales/pt-BR/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Intensidade de Raciocínio"
     },
+    "textVerbosity": {
+      "title": "Nível de detalhamento do texto de saída"
+    },
     "thinking": {
       "title": "Interruptor de Pensamento Profundo"
     },

package/locales/ru-RU/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Интенсивность рассуждений"
     },
+    "textVerbosity": {
+      "title": "Уровень подробности выводимого текста"
+    },
     "thinking": {
       "title": "Переключатель глубокого мышления"
     },

package/locales/tr-TR/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Akıl Yürütme Gücü"
     },
+    "textVerbosity": {
+      "title": "Çıktı metninin ayrıntı düzeyi"
+    },
     "thinking": {
       "title": "Derin Düşünme Anahtarı"
     },

package/locales/vi-VN/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "Cường độ suy luận"
     },
+    "textVerbosity": {
+      "title": "Mức độ chi tiết của văn bản đầu ra"
+    },
     "thinking": {
       "title": "Công tắc suy nghĩ sâu"
     },

package/locales/zh-CN/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "推理强度"
     },
+    "textVerbosity": {
+      "title": "输出文本详细程度"
+    },
     "thinking": {
       "title": "深度思考开关"
     },

package/locales/zh-TW/chat.json

@@ -46,6 +46,9 @@
     "reasoningEffort": {
       "title": "推理強度"
     },
+    "textVerbosity": {
+      "title": "輸出文本詳細程度"
+    },
     "thinking": {
       "title": "深度思考開關"
     },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lobehub/chat",
-  "version": "1.111.6",
+  "version": "1.111.8",
   "description": "Lobe Chat - an open-source, high-performance chatbot framework that supports speech synthesis, multimodal, and extensible Function Call plugin system. Supports one-click free deployment of your private ChatGPT/LLM web application.",
   "keywords": [
     "framework",
@@ -121,9 +121,8 @@
   "dependencies": {
     "@ant-design/icons": "^5.6.1",
     "@ant-design/pro-components": "^2.8.10",
-    "@anthropic-ai/sdk": "^0.57.0",
+    "@anthropic-ai/sdk": "^0.59.0",
     "@auth/core": "^0.40.0",
-    "@aws-sdk/client-bedrock-runtime": "^3.862.0",
     "@aws-sdk/client-s3": "^3.862.0",
     "@aws-sdk/s3-request-presigner": "^3.862.0",
     "@azure-rest/ai-inference": "1.0.0-beta.5",
@@ -145,6 +144,7 @@
     "@lobechat/electron-client-ipc": "workspace:*",
     "@lobechat/electron-server-ipc": "workspace:*",
     "@lobechat/file-loaders": "workspace:*",
+    "@lobechat/model-runtime": "workspace:*",
     "@lobechat/web-crawler": "workspace:*",
     "@lobehub/analytics": "^1.6.0",
     "@lobehub/charts": "^2.0.0",

package/packages/model-runtime/package.json

@@ -0,0 +1,12 @@
+{
+  "name": "@lobechat/model-runtime",
+  "version": "1.0.0",
+  "private": true,
+  "main": "./src/index.ts",
+  "dependencies": {
+    "@aws-sdk/client-bedrock-runtime": "^3.862.0",
+    "@lobechat/types": "workspace:*",
+    "debug": "^4.4.1",
+    "openai": "^4.104.0"
+  }
+}

package/{src/libs/model-runtime → packages/model-runtime/src}/anthropic/index.ts

@@ -204,14 +204,16 @@ export class LobeAnthropicAI implements LobeRuntimeAI {
       model,
       system: systemPrompts,
       // For Opus 4.1 models: prefer temperature over top_p if both are provided
-      temperature: isOpus41Model 
-        ? (shouldSetTemperature ? temperature / 2 : undefined)
-        : (payload.temperature !== undefined ? temperature / 2 : undefined),
+      temperature: isOpus41Model
+        ? shouldSetTemperature
+          ? temperature / 2
+          : undefined
+        : payload.temperature !== undefined
+          ? temperature / 2
+          : undefined,
       tools: postTools,
       // For Opus 4.1 models: only set top_p if temperature is not set
-      top_p: isOpus41Model 
-        ? (shouldSetTemperature ? undefined : top_p)
-        : top_p,
+      top_p: isOpus41Model ? (shouldSetTemperature ? undefined : top_p) : top_p,
     } satisfies Anthropic.MessageCreateParams;
   }
 

package/{src/libs/model-runtime → packages/model-runtime/src}/azureOpenai/index.ts

@@ -56,16 +56,33 @@ export class LobeAzureOpenAI implements LobeRuntimeAI {
     }));
 
     try {
-      const response = await this.client.chat.completions.create({
+      // Create parameters with proper typing for OpenAI SDK, handling reasoning_effort compatibility
+      const { reasoning_effort, ...otherParams } = params;
+
+      // Convert 'minimal' to 'low' for OpenAI SDK compatibility
+      const compatibleReasoningEffort = reasoning_effort === 'minimal' ? 'low' : reasoning_effort;
+
+      const baseParams = {
         messages: await convertOpenAIMessages(
           updatedMessages as OpenAI.ChatCompletionMessageParam[],
         ),
         model,
-        ...params,
+        ...otherParams,
         max_completion_tokens: undefined,
-        stream: enableStreaming,
-        tool_choice: params.tools ? 'auto' : undefined,
-      });
+        tool_choice: params.tools ? ('auto' as const) : undefined,
+      };
+
+      // Add reasoning_effort only if it exists and cast to proper type
+      const openaiParams = compatibleReasoningEffort
+        ? {
+            ...baseParams,
+            reasoning_effort: compatibleReasoningEffort as 'low' | 'medium' | 'high',
+          }
+        : baseParams;
+
+      const response = enableStreaming
+        ? await this.client.chat.completions.create({ ...openaiParams, stream: true })
+        : await this.client.chat.completions.create({ ...openaiParams, stream: false });
       if (enableStreaming) {
         const stream = response as Stream<OpenAI.ChatCompletionChunk>;
         const [prod, debug] = stream.tee();

package/{src/libs/model-runtime → packages/model-runtime/src}/google/index.ts

@@ -511,7 +511,7 @@ export class LobeGoogleAI implements LobeRuntimeAI {
       .map(async (msg) => await this.convertOAIMessagesToGoogleMessage(msg, toolCallNameMap));
 
     const contents = await Promise.all(pools);
-    
+
     // 筛除空消息: contents.parts must not be empty.
     return contents.filter((content: Content) => content.parts && content.parts.length > 0);
   };
@@ -563,7 +563,6 @@ export class LobeGoogleAI implements LobeRuntimeAI {
       },
     };
   };
-
 }
 
 export default LobeGoogleAI;

package/{src/libs/model-runtime → packages/model-runtime/src}/minimax/createImage.test.ts

@@ -55,22 +55,19 @@ describe('createMiniMaxImage', () => {
 
       const result = await createMiniMaxImage(payload, mockOptions);
 
-      expect(fetch).toHaveBeenCalledWith(
-        'https://api.minimaxi.com/v1/image_generation',
-        {
-          method: 'POST',
-          headers: {
-            'Authorization': 'Bearer test-api-key',
-            'Content-Type': 'application/json',
-          },
-          body: JSON.stringify({
-            aspect_ratio: undefined,
-            model: 'image-01',
-            n: 1,
-            prompt: 'A beautiful sunset over the mountains',
-          }),
+      expect(fetch).toHaveBeenCalledWith('https://api.minimaxi.com/v1/image_generation', {
+        method: 'POST',
+        headers: {
+          'Authorization': 'Bearer test-api-key',
+          'Content-Type': 'application/json',
         },
-      );
+        body: JSON.stringify({
+          aspect_ratio: undefined,
+          model: 'image-01',
+          n: 1,
+          prompt: 'A beautiful sunset over the mountains',
+        }),
+      });
 
       expect(result).toEqual({
         imageUrl: mockImageUrl,

package/{src/libs/model-runtime → packages/model-runtime/src}/minimax/createImage.ts

@@ -59,7 +59,7 @@ export async function createMiniMaxImage(
       }
 
       throw new Error(
-        `MiniMax API error (${response.status}): ${errorData?.base_resp || response.statusText}`
+        `MiniMax API error (${response.status}): ${errorData?.base_resp || response.statusText}`,
       );
     }
 
@@ -92,7 +92,6 @@ export async function createMiniMaxImage(
     log('Image generated successfully: %s', imageUrl);
 
     return { imageUrl };
-
   } catch (error) {
     log('Error in createMiniMaxImage: %O', error);
 

package/{src/libs/model-runtime → packages/model-runtime/src}/moonshot/index.ts

@@ -14,7 +14,7 @@ export const LobeMoonshotAI = createOpenAICompatibleRuntime({
       const { enabledSearch, messages, temperature, tools, ...rest } = payload;
 
       // 为 assistant 空消息添加一个空格 (#8418)
-      const filteredMessages = messages.map(message => {
+      const filteredMessages = messages.map((message) => {
         if (message.role === 'assistant' && (!message.content || message.content === '')) {
           return { ...message, content: ' ' };
         }

package/{src/libs/model-runtime → packages/model-runtime/src}/openai/index.ts

@@ -60,7 +60,7 @@ export const LobeOpenAI = createOpenAICompatibleRuntime({
   provider: ModelProvider.OpenAI,
   responses: {
     handlePayload: (payload) => {
-      const { enabledSearch, model, tools, ...rest } = payload;
+      const { enabledSearch, model, tools, verbosity, ...rest } = payload;
 
       const openaiTools = enabledSearch
         ? [
@@ -85,6 +85,9 @@ export const LobeOpenAI = createOpenAICompatibleRuntime({
           tools: openaiTools as any,
           // computer-use series must set truncation as auto
           ...(model.startsWith('computer-use') && { truncation: 'auto' }),
+          text: verbosity
+            ? { verbosity }
+            : undefined,
         }) as any;
       }
 

package/{src/libs/model-runtime → packages/model-runtime/src}/qwen/createImage.ts

@@ -24,7 +24,7 @@ const QwenText2ImageModels = [
   'wanx2.0-t2i',
   'wanx-v1',
   'flux',
-  'stable-diffusion'
+  'stable-diffusion',
 ];
 
 const getModelType = (model: string): string => {
@@ -33,12 +33,12 @@ const getModelType = (model: string): string => {
   //   return 'image2image';
   // }
 
-  if (QwenText2ImageModels.some(prefix => model.startsWith(prefix))) {
+  if (QwenText2ImageModels.some((prefix) => model.startsWith(prefix))) {
     return 'text2image';
   }
 
   throw new Error(`Unsupported model: ${model}`);
-}
+};
 
 /**
  * Create an image generation task with Qwen API
@@ -47,7 +47,7 @@ async function createImageTask(payload: CreateImagePayload, apiKey: string): Pro
   const { model, params } = payload;
   // I can only say that the design of Alibaba Cloud's API is really bad; each model has a different endpoint path.
   const modelType = getModelType(model);
-  const endpoint = `https://dashscope.aliyuncs.com/api/v1/services/aigc/${modelType}/image-synthesis`
+  const endpoint = `https://dashscope.aliyuncs.com/api/v1/services/aigc/${modelType}/image-synthesis`;
   if (!endpoint) {
     throw new Error(`No endpoint configured for model type: ${modelType}`);
   }
@@ -66,7 +66,7 @@ async function createImageTask(payload: CreateImagePayload, apiKey: string): Pro
         ...(typeof params.seed === 'number' ? { seed: params.seed } : {}),
         ...(params.width && params.height
           ? { size: `${params.width}*${params.height}` }
-          : params.size 
+          : params.size
             ? { size: params.size.replaceAll('x', '*') }
             : { size: '1024*1024' }),
       },

package/{src/libs/model-runtime → packages/model-runtime/src}/qwen/index.ts

@@ -36,8 +36,8 @@ export const LobeQwenAI = createOpenAICompatibleRuntime({
                 thinking?.budget_tokens === 0 ? 0 : thinking?.budget_tokens || undefined,
             }
           : ['qwen3', 'qwen-turbo', 'qwen-plus'].some((keyword) =>
-              model.toLowerCase().includes(keyword),
-            )
+                model.toLowerCase().includes(keyword),
+              )
             ? {
                 enable_thinking: thinking !== undefined ? thinking.type === 'enabled' : false,
                 thinking_budget:

package/{src/libs/model-runtime → packages/model-runtime/src}/sensenova/index.ts

@@ -12,8 +12,16 @@ export const LobeSenseNovaAI = createOpenAICompatibleRuntime({
   baseURL: 'https://api.sensenova.cn/compatible-mode/v1',
   chatCompletion: {
     handlePayload: (payload) => {
-      const { frequency_penalty, max_tokens, messages, model, temperature, thinking, top_p, ...rest } =
-        payload;
+      const {
+        frequency_penalty,
+        max_tokens,
+        messages,
+        model,
+        temperature,
+        thinking,
+        top_p,
+        ...rest
+      } = payload;
 
       return {
         ...rest,
@@ -33,8 +41,10 @@ export const LobeSenseNovaAI = createOpenAICompatibleRuntime({
           temperature !== undefined && temperature > 0 && temperature <= 2
             ? temperature
             : undefined,
-        thinking: thinking 
-          ? (model && model.includes('-V6-5-') && thinking.type === 'enabled' ? { enabled: true } : { enabled: false })
+        thinking: thinking
+          ? model && model.includes('-V6-5-') && thinking.type === 'enabled'
+            ? { enabled: true }
+            : { enabled: false }
           : undefined,
         top_p: top_p !== undefined && top_p > 0 && top_p < 1 ? top_p : undefined,
       } as any;

package/{src/libs/model-runtime → packages/model-runtime/src}/types/chat.ts

@@ -107,7 +107,7 @@ export interface ChatStreamPayload {
     effort?: string;
     summary?: string;
   };
-  reasoning_effort?: 'low' | 'medium' | 'high';
+  reasoning_effort?: 'minimal' | 'low' | 'medium' | 'high';
   responseMode?: 'stream' | 'json';
   /**
    * @title 是否开启流式请求
@@ -119,6 +119,9 @@ export interface ChatStreamPayload {
    * @default 1
    */
   temperature: number;
+  text?: {
+    verbosity?: 'low' | 'medium' | 'high';
+  };
   /**
    * use for Claude and Gemini
    */
@@ -135,6 +138,7 @@ export interface ChatStreamPayload {
    */
   top_p?: number;
   truncation?: 'auto' | 'disabled';
+  verbosity?: 'low' | 'medium' | 'high';
 }
 
 export interface ChatMethodOptions {