Add latest models and remove legacy ones

.github/workflows/test.yml

@@ -21,19 +21,19 @@ jobs:
       - name: Set up Node.js
         uses: actions/setup-node@v3
         with:
-          node-version: 18
-          cache: "yarn"
+          node-version: 23
+          cache: 'npm'
 
       - name: Cache node_modules
         uses: actions/cache@v4
         with:
           path: node_modules
-          key: ${{ runner.os }}-node_modules-${{ hashFiles('**/yarn.lock') }}
+          key: ${{ runner.os }}-node_modules-${{ hashFiles('**/package-lock.json') }}
           restore-keys: |
             ${{ runner.os }}-node_modules-
 
       - name: Install dependencies
-        run: yarn install
+        run: npm install
 
       - name: Run Jest tests
-        run: yarn test:ci
+        run: npm run test:ci

.gitignore

@@ -49,3 +49,4 @@ masks.json
 
 # mcp config
 app/mcp/mcp_config.json
+Dockerfile.local

Dockerfile

@@ -1,29 +1,25 @@
-FROM node:18-alpine AS base
+FROM node:23-alpine AS base
 
 FROM base AS deps
 
 RUN apk add --no-cache libc6-compat
 
 WORKDIR /app
 
-COPY package.json yarn.lock ./
+COPY package.json package-lock.json ./
 
-RUN yarn config set registry 'https://registry.npmmirror.com/'
-RUN yarn install
+RUN npm config set registry 'https://registry.npmmirror.com/'
+RUN npm install
 
 FROM base AS builder
 
 RUN apk update && apk add --no-cache git
 
-ENV OPENAI_API_KEY=""
-ENV GOOGLE_API_KEY=""
-ENV CODE=""
-
 WORKDIR /app
 COPY --from=deps /app/node_modules ./node_modules
 COPY . .
 
-RUN yarn build
+RUN npm run build
 
 FROM base AS runner
 WORKDIR /app
@@ -33,6 +29,8 @@ RUN apk add proxychains-ng
 ENV PROXY_URL=""
 ENV OPENAI_API_KEY=""
 ENV GOOGLE_API_KEY=""
+ENV ANTHROPIC_API_KEY=""
+ENV VISION_MODELS=""
 ENV CODE=""
 ENV ENABLE_MCP=""
 

app/client/platforms/anthropic.ts

@@ -71,8 +71,6 @@ const ClaudeMapper = {
   system: "user",
 } as const;
 
-const keys = ["claude-2, claude-instant-1"];
-
 export class ClaudeApi implements LLMApi {
   speech(options: SpeechOptions): Promise<ArrayBuffer> {
     throw new Error("Method not implemented.");

app/client/platforms/google.ts

@@ -197,8 +197,6 @@ export class GeminiProApi implements LLMApi {
         signal: controller.signal,
         headers: getHeaders(),
       };
-
-      const isThinking = options.config.model.includes("-thinking");
       // make a fetch request
       const requestTimeoutId = setTimeout(
         () => controller.abort(),

app/client/platforms/openai.ts

@@ -67,6 +67,8 @@ export interface RequestPayload {
   top_p: number;
   max_tokens?: number;
   max_completion_tokens?: number;
+  reasoning_effort?: string;
+  // O3 only
 }
 
 export interface DalleRequestPayload {
@@ -196,9 +198,10 @@ export class ChatGPTApi implements LLMApi {
     let requestPayload: RequestPayload | DalleRequestPayload;
 
     const isDalle3 = _isDalle3(options.config.model);
-    const isO1OrO3 =
-      options.config.model.startsWith("o1") ||
-      options.config.model.startsWith("o3");
+    const isO1 = options.config.model.startsWith("o1");
+    const isO3 = options.config.model.startsWith("o3");
+    const isO4 = options.config.model.startsWith("o4");
+    const isO1OrO3orO4 = isO1 || isO3 || isO4;
     if (isDalle3) {
       const prompt = getMessageTextContent(
         options.messages.slice(-1)?.pop() as any,
@@ -220,7 +223,7 @@ export class ChatGPTApi implements LLMApi {
         const content = visionModel
           ? await preProcessImageContent(v.content)
           : getMessageTextContent(v);
-        if (!(isO1OrO3 && v.role === "system"))
+        if (!(isO1OrO3orO4 && v.role === "system"))
           messages.push({ role: v.role, content });
       }
 
@@ -229,22 +232,31 @@ export class ChatGPTApi implements LLMApi {
         messages,
         stream: options.config.stream,
         model: modelConfig.model,
-        temperature: !isO1OrO3 ? modelConfig.temperature : 1,
-        presence_penalty: !isO1OrO3 ? modelConfig.presence_penalty : 0,
-        frequency_penalty: !isO1OrO3 ? modelConfig.frequency_penalty : 0,
-        top_p: !isO1OrO3 ? modelConfig.top_p : 1,
+        temperature: !isO1OrO3orO4 ? modelConfig.temperature : 1,
+        presence_penalty: !isO1OrO3orO4 ? modelConfig.presence_penalty : 0,
+        frequency_penalty: !isO1OrO3orO4 ? modelConfig.frequency_penalty : 0,
+        top_p: !isO1OrO3orO4 ? modelConfig.top_p : 1,
         // max_tokens: Math.max(modelConfig.max_tokens, 1024),
         // Please do not ask me why not send max_tokens, no reason, this param is just shit, I dont want to explain anymore.
       };
 
       // O1 使用 max_completion_tokens 控制token数 (https://platform.openai.com/docs/guides/reasoning#controlling-costs)
-      if (isO1OrO3) {
-        requestPayload["max_completion_tokens"] = modelConfig.max_tokens;
+      if (isO1OrO3orO4) {
+        requestPayload["max_completion_tokens"] = 20000;
+      }
+
+      if (isO4) {
+        requestPayload["reasoning_effort"] = "high";
+        // make o4-mini defaults to high reasoning effort
       }
 
       // add max_tokens to vision model
       if (visionModel) {
-        requestPayload["max_tokens"] = Math.max(modelConfig.max_tokens, 4000);
+        if (isO1OrO3orO4) {
+          requestPayload["max_completion_tokens"] = 20000;
+        } else {
+          requestPayload["max_tokens"] = Math.max(modelConfig.max_tokens, 4000);
+        }
       }
     }
 
@@ -286,6 +298,11 @@ export class ChatGPTApi implements LLMApi {
           isDalle3 ? OpenaiPath.ImagePath : OpenaiPath.ChatPath,
         );
       }
+      // make a fetch request
+      const requestTimeoutId = setTimeout(
+        () => controller.abort(),
+        getTimeoutMSByModel(options.config.model),
+      );
       if (shouldStream) {
         let index = -1;
         const [tools, funcs] = usePluginStore
@@ -393,12 +410,6 @@ export class ChatGPTApi implements LLMApi {
           headers: getHeaders(),
         };
 
-        // make a fetch request
-        const requestTimeoutId = setTimeout(
-          () => controller.abort(),
-          getTimeoutMSByModel(options.config.model),
-        );
-
         const res = await fetch(chatPath, chatPayload);
         clearTimeout(requestTimeoutId);