@lobehub/chat 1.94.4 → 1.94.6

package/CHANGELOG.md

@@ -2,6 +2,64 @@
 
 # Changelog
 
+### [Version 1.94.6](https://github.com/lobehub/lobe-chat/compare/v1.94.5...v1.94.6)
+
+<sup>Released on **2025-06-12**</sup>
+
+#### 🐛 Bug Fixes
+
+- **misc**: Abort the Gemini request correctly & Add openai o3-pro.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### What's fixed
+
+- **misc**: Abort the Gemini request correctly & Add openai o3-pro, closes [#8135](https://github.com/lobehub/lobe-chat/issues/8135) ([c79f1b9](https://github.com/lobehub/lobe-chat/commit/c79f1b9))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
+### [Version 1.94.5](https://github.com/lobehub/lobe-chat/compare/v1.94.4...v1.94.5)
+
+<sup>Released on **2025-06-12**</sup>
+
+#### 🐛 Bug Fixes
+
+- **chat**: Improve response animation merging logic.
+
+#### 💄 Styles
+
+- **misc**: Support `web_search_preview` & fix some bug form OpenAI Response API.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### What's fixed
+
+- **chat**: Improve response animation merging logic, closes [#8160](https://github.com/lobehub/lobe-chat/issues/8160) ([9d81cdc](https://github.com/lobehub/lobe-chat/commit/9d81cdc))
+
+#### Styles
+
+- **misc**: Support `web_search_preview` & fix some bug form OpenAI Response API, closes [#8131](https://github.com/lobehub/lobe-chat/issues/8131) ([b2983f0](https://github.com/lobehub/lobe-chat/commit/b2983f0))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
 ### [Version 1.94.4](https://github.com/lobehub/lobe-chat/compare/v1.94.3...v1.94.4)
 
 <sup>Released on **2025-06-11**</sup>

package/README.md

@@ -335,7 +335,7 @@ In addition, these plugins are not limited to news aggregation, but can also ext
 | [Bing_websearch](https://lobechat.com/discover/plugin/Bingsearch-identifier)<br/><sup>By **FineHow** on **2024-12-22**</sup> | Search for information from the internet base BingApi<br/>`bingsearch`                                                  |
 | [Google CSE](https://lobechat.com/discover/plugin/google-cse)<br/><sup>By **vsnthdev** on **2024-12-02**</sup>               | Searches Google through their official CSE API.<br/>`web` `search`                                                      |
 
-> 📊 Total plugins: [<kbd>**43**</kbd>](https://lobechat.com/discover/plugins)
+> 📊 Total plugins: [<kbd>**42**</kbd>](https://lobechat.com/discover/plugins)
 
  <!-- PLUGIN LIST -->
 

package/README.zh-CN.md

@@ -328,7 +328,7 @@ LobeChat 的插件生态系统是其核心功能的重要扩展，它极大地
 | [必应网页搜索](https://lobechat.com/discover/plugin/Bingsearch-identifier)<br/><sup>By **FineHow** on **2024-12-22**</sup> | 通过 BingApi 搜索互联网上的信息<br/>`bingsearch`                                   |
 | [谷歌自定义搜索引擎](https://lobechat.com/discover/plugin/google-cse)<br/><sup>By **vsnthdev** on **2024-12-02**</sup>     | 通过他们的官方自定义搜索引擎 API 搜索谷歌。<br/>`网络` `搜索`                      |
 
-> 📊 Total plugins: [<kbd>**43**</kbd>](https://lobechat.com/discover/plugins)
+> 📊 Total plugins: [<kbd>**42**</kbd>](https://lobechat.com/discover/plugins)
 
  <!-- PLUGIN LIST -->
 

package/changelog/v1.json

@@ -1,4 +1,22 @@
 [
+  {
+    "children": {
+      "fixes": [
+        "Abort the Gemini request correctly & Add openai o3-pro."
+      ]
+    },
+    "date": "2025-06-12",
+    "version": "1.94.6"
+  },
+  {
+    "children": {
+      "improvements": [
+        "Support web_search_preview & fix some bug form OpenAI Response API."
+      ]
+    },
+    "date": "2025-06-12",
+    "version": "1.94.5"
+  },
   {
     "children": {
       "improvements": [

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lobehub/chat",
-  "version": "1.94.4",
+  "version": "1.94.6",
   "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",
@@ -54,7 +54,7 @@
     "dev:desktop": "next dev --turbopack -p 3015",
     "docs:i18n": "lobe-i18n md && npm run lint:md && npm run lint:mdx && prettier -c --write locales/**/*",
     "docs:seo": "lobe-seo && npm run lint:mdx",
-    "i18n": "npm run workflow:i18n && lobe-i18n",
+    "i18n": "npm run workflow:i18n && lobe-i18n && prettier -c --write \"locales/**\"",
     "lint": "npm run lint:ts && npm run lint:style && npm run type-check && npm run lint:circular",
     "lint:circular": "dpdm src/**/*.ts  --no-warning --no-tree --exit-code circular:1 --no-progress -T true --skip-dynamic-imports circular",
     "lint:md": "remark . --silent --output",
@@ -282,6 +282,7 @@
     "@next/bundle-analyzer": "^15.3.3",
     "@next/eslint-plugin-next": "^15.3.3",
     "@peculiar/webcrypto": "^1.5.0",
+    "@prettier/sync": "^0.6.1",
     "@semantic-release/exec": "^6.0.3",
     "@testing-library/jest-dom": "^6.6.3",
     "@testing-library/react": "^16.3.0",

package/scripts/i18nWorkflow/genDefaultLocale.ts

@@ -2,7 +2,7 @@ import { consola } from 'consola';
 import { colors } from 'consola/utils';
 
 import { entryLocaleJsonFilepath, i18nConfig, srcDefaultLocales } from './const';
-import { tagWhite, writeJSON } from './utils';
+import { tagWhite, writeJSONWithPrettier } from './utils';
 
 export const genDefaultLocale = () => {
   consola.info(`Default locale is ${i18nConfig.entryLocale}...`);
@@ -13,7 +13,7 @@ export const genDefaultLocale = () => {
 
   for (const [ns, value] of data) {
     const filepath = entryLocaleJsonFilepath(`${ns}.json`);
-    writeJSON(filepath, value);
+    writeJSONWithPrettier(filepath, value);
     consola.success(tagWhite(ns), colors.gray(filepath));
   }
 };

package/scripts/i18nWorkflow/genDiff.ts

@@ -10,10 +10,10 @@ import {
   outputLocaleJsonFilepath,
   srcDefaultLocales,
 } from './const';
-import { readJSON, tagWhite, writeJSON } from './utils';
+import { readJSON, tagWhite, writeJSONWithPrettier } from './utils';
 
 export const genDiff = () => {
-  consola.start(`Diff between Dev/Prod local...`);
+  consola.start(`Remove diff analysis...`);
 
   const resources = require(srcDefaultLocales);
   const data = Object.entries(resources.default);
@@ -21,27 +21,26 @@ export const genDiff = () => {
   for (const [ns, devJSON] of data) {
     const filepath = entryLocaleJsonFilepath(`${ns}.json`);
     if (!existsSync(filepath)) continue;
-    const prodJSON = readJSON(filepath);
+    const previousProdJSON = readJSON(filepath);
 
-    const diffResult = diff(prodJSON, devJSON as any);
-    const remove = diffResult.filter((item) => item.op === 'remove');
-    if (remove.length === 0) {
+    const diffResult = diff(previousProdJSON, devJSON as any);
+    if (diffResult.length === 0) {
       consola.success(tagWhite(ns), colors.gray(filepath));
       continue;
     }
 
     const clearLocals = [];
 
-    for (const locale of [i18nConfig.entryLocale, ...i18nConfig.outputLocales]) {
+    for (const locale of i18nConfig.outputLocales) {
       const localeFilepath = outputLocaleJsonFilepath(locale, `${ns}.json`);
       if (!existsSync(localeFilepath)) continue;
       const localeJSON = readJSON(localeFilepath);
 
-      for (const item of remove) {
+      for (const item of diffResult) {
         unset(localeJSON, item.path);
       }
 
-      writeJSON(localeFilepath, localeJSON);
+      writeJSONWithPrettier(localeFilepath, localeJSON);
       clearLocals.push(locale);
     }
     consola.info('clear', clearLocals);

package/scripts/i18nWorkflow/utils.ts

@@ -2,9 +2,13 @@ import { consola } from 'consola';
 import { colors } from 'consola/utils';
 import { readFileSync, writeFileSync } from 'node:fs';
 import { resolve } from 'node:path';
-
+import prettier from "@prettier/sync";
 import i18nConfig from '../../.i18nrc';
 
+let prettierOptions = prettier.resolveConfig(
+  resolve(__dirname, '../../.prettierrc.js')
+);
+
 export const readJSON = (filePath: string) => {
   const data = readFileSync(filePath, 'utf8');
   return JSON.parse(data);
@@ -15,6 +19,15 @@ export const writeJSON = (filePath: string, data: any) => {
   writeFileSync(filePath, jsonStr, 'utf8');
 };
 
+export const writeJSONWithPrettier = (filePath: string, data: any) => {
+  const jsonStr = JSON.stringify(data, null, 2);
+  const formatted = prettier.format(jsonStr, {
+    ...prettierOptions,
+    parser: 'json',
+  });
+  writeFileSync(filePath, formatted, 'utf8');
+};
+
 export const genResourcesContent = (locales: string[]) => {
   let index = '';
   let indexObj = '';

package/src/config/aiModels/groq.ts

@@ -62,6 +62,20 @@ const groqChatModels: AIChatModelCard[] = [
     },
     type: 'chat',
   },
+  {
+    abilities: {
+      reasoning: true,
+    },
+    contextWindowTokens: 131_072,
+    displayName: 'Qwen3 32B',
+    id: 'qwen/qwen3-32b',
+    maxOutput: 16_384,
+    pricing: {
+      input: 0.29,
+      output: 0.59,
+    },
+    type: 'chat',
+  },
   {
     abilities: {
       functionCall: true,

package/src/config/aiModels/openai.ts

@@ -8,6 +8,28 @@ import {
 } from '@/types/aiModel';
 
 export const openaiChatModels: AIChatModelCard[] = [
+  {
+    abilities: {
+      functionCall: true,
+      reasoning: true,
+      vision: true,
+    },
+    contextWindowTokens: 200_000,
+    description:
+      'o3-pro 模型使用更多的计算来更深入地思考并始终提供更好的答案，仅支持 Responses API 下使用。',
+    displayName: 'o3-pro',
+    id: 'o3-pro',
+    maxOutput: 100_000,
+    pricing: {
+      input: 20,
+      output: 80,
+    },
+    releasedAt: '2025-06-10',
+    settings: {
+      extendParams: ['reasoningEffort'],
+    },
+    type: 'chat',
+  },
   {
     abilities: {
       functionCall: true,
@@ -22,11 +44,11 @@ export const openaiChatModels: AIChatModelCard[] = [
     id: 'o3',
     maxOutput: 100_000,
     pricing: {
-      cachedInput: 2.5,
-      input: 10,
-      output: 40,
+      cachedInput: 0.5,
+      input: 2,
+      output: 8,
     },
-    releasedAt: '2025-04-17',
+    releasedAt: '2025-04-16',
     settings: {
       extendParams: ['reasoningEffort'],
     },
@@ -59,6 +81,7 @@ export const openaiChatModels: AIChatModelCard[] = [
   {
     abilities: {
       functionCall: true,
+      search: true,
       vision: true,
     },
     contextWindowTokens: 1_047_576,
@@ -73,11 +96,15 @@ export const openaiChatModels: AIChatModelCard[] = [
       output: 8,
     },
     releasedAt: '2025-04-14',
+    settings: {
+      searchImpl: 'params',
+    },
     type: 'chat',
   },
   {
     abilities: {
       functionCall: true,
+      search: true,
       vision: true,
     },
     contextWindowTokens: 1_047_576,
@@ -93,6 +120,9 @@ export const openaiChatModels: AIChatModelCard[] = [
       output: 1.6,
     },
     releasedAt: '2025-04-14',
+    settings: {
+      searchImpl: 'params',
+    },
     type: 'chat',
   },
   {
@@ -135,6 +165,28 @@ export const openaiChatModels: AIChatModelCard[] = [
     },
     type: 'chat',
   },
+  {
+    abilities: {
+      functionCall: true,
+      reasoning: true,
+      vision: true,
+    },
+    contextWindowTokens: 200_000,
+    description:
+      'o1 系列模型经过强化学习训练，能够在回答前进行思考，并执行复杂的推理任务。o1-pro 模型使用了更多计算资源，以进行更深入的思考，从而持续提供更优质的回答。',
+    displayName: 'o1-pro',
+    id: 'o1-pro',
+    maxOutput: 100_000,
+    pricing: {
+      input: 150,
+      output: 600,
+    },
+    releasedAt: '2025-03-19',
+    settings: {
+      extendParams: ['reasoningEffort'],
+    },
+    type: 'chat',
+  },
   {
     abilities: {
       reasoning: true,
@@ -158,6 +210,7 @@ export const openaiChatModels: AIChatModelCard[] = [
   },
   {
     abilities: {
+      functionCall: true,
       reasoning: true,
       vision: true,
     },
@@ -220,6 +273,7 @@ export const openaiChatModels: AIChatModelCard[] = [
   {
     abilities: {
       functionCall: true,
+      search: true,
       vision: true,
     },
     contextWindowTokens: 128_000,
@@ -234,6 +288,9 @@ export const openaiChatModels: AIChatModelCard[] = [
       output: 0.6,
     },
     releasedAt: '2024-07-18',
+    settings: {
+      searchImpl: 'params',
+    },
     type: 'chat',
   },
   {
@@ -259,6 +316,29 @@ export const openaiChatModels: AIChatModelCard[] = [
   {
     abilities: {
       functionCall: true,
+      //search: true,
+    },
+    contextWindowTokens: 128_000,
+    description: 'GPT-4o mini Audio 模型，支持音频输入输出',
+    displayName: 'GPT-4o mini Audio',
+    id: 'gpt-4o-mini-audio-preview',
+    maxOutput: 16_384,
+    pricing: {
+      input: 0.15,
+      output: 0.6,
+    },
+    releasedAt: '2024-12-17',
+    /*
+    settings: {
+      searchImpl: 'params',
+    },
+    */
+    type: 'chat',
+  },
+  {
+    abilities: {
+      functionCall: true,
+      search: true,
       vision: true,
     },
     contextWindowTokens: 128_000,
@@ -272,6 +352,9 @@ export const openaiChatModels: AIChatModelCard[] = [
       output: 10,
     },
     releasedAt: '2024-05-13',
+    settings: {
+      searchImpl: 'params',
+    },
     type: 'chat',
   },
   {
@@ -297,6 +380,7 @@ export const openaiChatModels: AIChatModelCard[] = [
   {
     abilities: {
       functionCall: true,
+      search: true,
       vision: true,
     },
     contextWindowTokens: 128_000,
@@ -310,11 +394,15 @@ export const openaiChatModels: AIChatModelCard[] = [
       output: 10,
     },
     releasedAt: '2024-11-20',
+    settings: {
+      searchImpl: 'params',
+    },
     type: 'chat',
   },
   {
     abilities: {
       functionCall: true,
+      search: true,
       vision: true,
     },
     contextWindowTokens: 128_000,
@@ -327,9 +415,16 @@ export const openaiChatModels: AIChatModelCard[] = [
       output: 15,
     },
     releasedAt: '2024-05-13',
+    settings: {
+      searchImpl: 'params',
+    },
     type: 'chat',
   },
   {
+    abilities: {
+      functionCall: true,
+      //search: true,
+    },
     contextWindowTokens: 128_000,
     description: 'GPT-4o Audio 模型，支持音频输入输出',
     displayName: 'GPT-4o Audio',
@@ -340,6 +435,11 @@ export const openaiChatModels: AIChatModelCard[] = [
       output: 10,
     },
     releasedAt: '2024-10-01',
+    /*
+    settings: {
+      searchImpl: 'params',
+    },
+    */
     type: 'chat',
   },
   {
@@ -545,6 +645,48 @@ export const openaiChatModels: AIChatModelCard[] = [
     },
     type: 'chat',
   },
+  {
+    abilities: {
+      functionCall: true,
+      reasoning: true,
+      vision: true,
+    },
+    contextWindowTokens: 200_000,
+    description: 'codex-mini-latest 是 o4-mini 的微调版本，专门用于 Codex CLI。对于直接通过 API 使用，我们推荐从 gpt-4.1 开始。',
+    displayName: 'Codex mini',
+    id: 'codex-mini-latest',
+    maxOutput: 100_000,
+    pricing: {
+      input: 1.5,
+      output: 6,
+    },
+    releasedAt: '2025-06-01',
+    settings: {
+      extendParams: ['reasoningEffort'],
+    },
+    type: 'chat',
+  },
+  {
+    abilities: {
+      functionCall: true,
+      reasoning: true,
+      vision: true,
+    },
+    contextWindowTokens: 8192,
+    description: 'computer-use-preview 模型是专为“计算机使用工具”设计的专用模型，经过训练以理解并执行计算机相关任务。',
+    displayName: 'Computer Use Preview',
+    id: 'computer-use-preview',
+    maxOutput: 1024,
+    pricing: {
+      input: 3,
+      output: 12,
+    },
+    releasedAt: '2025-03-11',
+    settings: {
+      extendParams: ['reasoningEffort'],
+    },
+    type: 'chat',
+  },
 ];
 
 export const openaiEmbeddingModels: AIEmbeddingModelCard[] = [

package/src/config/modelProviders/openai.ts

@@ -323,7 +323,7 @@ const OpenAI: ModelProviderCard = {
       },
     },
   ],
-  checkModel: 'gpt-4o-mini',
+  checkModel: 'gpt-4.1-nano',
   description:
     'OpenAI 是全球领先的人工智能研究机构，其开发的模型如GPT系列推动了自然语言处理的前沿。OpenAI 致力于通过创新和高效的AI解决方案改变多个行业。他们的产品具有显著的性能和经济性，广泛用于研究、商业和创新应用。',
   enabled: true,

package/src/const/models.ts

@@ -6,7 +6,32 @@ export const systemToUserModels = new Set([
 ]);
 
 // TODO: 临时写法，后续要重构成 model card 展示配置
-export const disableStreamModels = new Set(['o1', 'o1-2024-12-17']);
+export const disableStreamModels = new Set([
+  'o1',
+  'o1-2024-12-17',
+  'o1-pro',
+  'o1-pro-2025-03-19',
+  /*
+  官网显示不支持，但是实际试下来支持 Streaming，暂时注释掉
+  'o3-pro',
+  'o3-pro-2025-06-10',
+  */
+  'computer-use-preview',
+  'computer-use-preview-2025-03-11',
+]);
+
+/**
+ * models use Responses API only
+ */
+export const responsesAPIModels = new Set([
+  'o1-pro',
+  'o1-pro-2025-03-19',
+  'o3-pro',
+  'o3-pro-2025-06-10',
+  'codex-mini-latest',
+  'computer-use-preview',
+  'computer-use-preview-2025-03-11',
+]);
 
 /**
  * models support context caching

package/src/libs/model-runtime/google/index.ts

@@ -25,11 +25,7 @@ import {
 import { AgentRuntimeError } from '../utils/createError';
 import { debugStream } from '../utils/debugStream';
 import { StreamingResponse } from '../utils/response';
-import {
-  GoogleGenerativeAIStream,
-  VertexAIStream,
-  convertIterableToStream,
-} from '../utils/streams';
+import { GoogleGenerativeAIStream, VertexAIStream } from '../utils/streams';
 import { parseDataUri } from '../utils/uriParser';
 
 const modelsOffSafetySettings = new Set(['gemini-2.0-flash-exp']);
@@ -91,6 +87,17 @@ interface GoogleAIThinkingConfig {
   thinkingBudget?: number;
 }
 
+const isAbortError = (error: Error): boolean => {
+  const message = error.message.toLowerCase();
+  return (
+    message.includes('aborted') ||
+    message.includes('cancelled') ||
+    message.includes('error reading from the stream') ||
+    message.includes('abort') ||
+    error.name === 'AbortError'
+  );
+};
+
 export class LobeGoogleAI implements LobeRuntimeAI {
   private client: GoogleGenerativeAI;
   private isVertexAi: boolean;
@@ -140,6 +147,20 @@ export class LobeGoogleAI implements LobeRuntimeAI {
       const contents = await this.buildGoogleMessages(payload.messages);
 
       const inputStartAt = Date.now();
+
+      const controller = new AbortController();
+      const originalSignal = options?.signal;
+
+      if (originalSignal) {
+        if (originalSignal.aborted) {
+          controller.abort();
+        } else {
+          originalSignal.addEventListener('abort', () => {
+            controller.abort();
+          });
+        }
+      }
+
       const geminiStreamResult = await this.client
         .getGenerativeModel(
           {
@@ -177,15 +198,20 @@ export class LobeGoogleAI implements LobeRuntimeAI {
           },
           { apiVersion: 'v1beta', baseUrl: this.baseURL },
         )
-        .generateContentStream({
-          contents,
-          systemInstruction: modelsDisableInstuction.has(model)
-            ? undefined
-            : (payload.system as string),
-          tools: this.buildGoogleTools(payload.tools, payload),
-        });
+        .generateContentStream(
+          {
+            contents,
+            systemInstruction: modelsDisableInstuction.has(model)
+              ? undefined
+              : (payload.system as string),
+            tools: this.buildGoogleTools(payload.tools, payload),
+          },
+          {
+            signal: controller.signal,
+          },
+        );
 
-      const googleStream = convertIterableToStream(geminiStreamResult.stream);
+      const googleStream = this.createEnhancedStream(geminiStreamResult.stream, controller.signal);
       const [prod, useForDebug] = googleStream.tee();
 
       const key = this.isVertexAi
@@ -205,6 +231,16 @@ export class LobeGoogleAI implements LobeRuntimeAI {
     } catch (e) {
       const err = e as Error;
 
+      // 移除之前的静默处理，统一抛出错误
+      if (isAbortError(err)) {
+        console.log('Request was cancelled');
+        throw AgentRuntimeError.chat({
+          error: { message: 'Request was cancelled' },
+          errorType: AgentRuntimeErrorType.ProviderBizError,
+          provider: this.provider,
+        });
+      }
+
       console.log(err);
       const { errorType, error } = this.parseErrorMessage(err.message);
 
@@ -212,24 +248,75 @@ export class LobeGoogleAI implements LobeRuntimeAI {
     }
   }
 
-  async models() {
+  private createEnhancedStream(originalStream: any, signal: AbortSignal): ReadableStream {
+    return new ReadableStream({
+      async start(controller) {
+        let hasData = false;
+
+        try {
+          for await (const chunk of originalStream) {
+            if (signal.aborted) {
+              // 如果有数据已经输出，优雅地关闭流而不是抛出错误
+              if (hasData) {
+                console.log('Stream cancelled gracefully, preserving existing output');
+                controller.close();
+                return;
+              } else {
+                // 如果还没有数据输出，则抛出取消错误
+                throw new Error('Stream cancelled');
+              }
+            }
+
+            hasData = true;
+            controller.enqueue(chunk);
+          }
+        } catch (error) {
+          const err = error as Error;
+
+          // 统一处理所有错误，包括 abort 错误
+          if (isAbortError(err) || signal.aborted) {
+            // 如果有数据已经输出，优雅地关闭流
+            if (hasData) {
+              console.log('Stream reading cancelled gracefully, preserving existing output');
+              controller.close();
+              return;
+            } else {
+              console.log('Stream reading cancelled before any output');
+              controller.error(new Error('Stream cancelled'));
+              return;
+            }
+          } else {
+            // 处理其他流解析错误
+            console.error('Stream parsing error:', err);
+            controller.error(err);
+            return;
+          }
+        }
+
+        controller.close();
+      },
+    });
+  }
+
+  async models(options?: { signal?: AbortSignal }) {
     try {
       const url = `${this.baseURL}/v1beta/models?key=${this.apiKey}`;
       const response = await fetch(url, {
         method: 'GET',
+        signal: options?.signal,
       });
-  
+
       if (!response.ok) {
         throw new Error(`HTTP error! status: ${response.status}`);
       }
-  
+
       const json = await response.json();
-      
+
       const modelList: GoogleModelCard[] = json.models;
-  
+
       const processedModels = modelList.map((model) => {
         const id = model.name.replace(/^models\//, '');
-        
+
         return {
           contextWindowTokens: (model.inputTokenLimit || 0) + (model.outputTokenLimit || 0),
           displayName: model.displayName || id,
@@ -237,9 +324,9 @@ export class LobeGoogleAI implements LobeRuntimeAI {
           maxOutput: model.outputTokenLimit || undefined,
         };
       });
-  
+
       const { MODEL_LIST_CONFIGS, processModelList } = await import('../utils/modelParse');
-      
+
       return processModelList(processedModels, MODEL_LIST_CONFIGS.google);
     } catch (error) {
       console.error('Failed to fetch Google models:', error);

package/src/libs/model-runtime/openai/index.ts

@@ -2,21 +2,24 @@ import { ChatStreamPayload, ModelProvider } from '../types';
 import { processMultiProviderModelList } from '../utils/modelParse';
 import { createOpenAICompatibleRuntime } from '../utils/openaiCompatibleFactory';
 import { pruneReasoningPayload } from '../utils/openaiHelpers';
+import { responsesAPIModels } from '@/const/models';
 
 export interface OpenAIModelCard {
   id: string;
 }
 
-const prunePrefixes = ['o1', 'o3', 'o4'];
+const prunePrefixes = ['o1', 'o3', 'o4', 'codex', 'computer-use'];
+
+const oaiSearchContextSize = process.env.OPENAI_SEARCH_CONTEXT_SIZE; // low, medium, high
 
 export const LobeOpenAI = createOpenAICompatibleRuntime({
   baseURL: 'https://api.openai.com/v1',
   chatCompletion: {
     handlePayload: (payload) => {
-      const { model } = payload;
+      const { enabledSearch, model, ...rest } = payload;
 
-      if (model === 'o1-pro') {
-        return { ...payload, apiMode: 'responses' } as ChatStreamPayload;
+      if (responsesAPIModels.has(model) || enabledSearch) {
+        return { ...rest, apiMode: 'responses', enabledSearch, model } as ChatStreamPayload;
       }
 
       if (prunePrefixes.some((prefix) => model.startsWith(prefix))) {
@@ -24,11 +27,10 @@ export const LobeOpenAI = createOpenAICompatibleRuntime({
       }
 
       if (model.includes('-search-')) {
-        const oaiSearchContextSize = process.env.OPENAI_SEARCH_CONTEXT_SIZE; // low, medium, high
-
         return {
-          ...payload,
+          ...rest,
           frequency_penalty: undefined,
+          model,
           presence_penalty: undefined,
           stream: payload.stream ?? true,
           temperature: undefined,
@@ -41,7 +43,7 @@ export const LobeOpenAI = createOpenAICompatibleRuntime({
         } as any;
       }
 
-      return { ...payload, stream: payload.stream ?? true };
+      return { ...rest, model, stream: payload.stream ?? true };
     },
   },
   debug: {
@@ -57,17 +59,37 @@ export const LobeOpenAI = createOpenAICompatibleRuntime({
   },
   provider: ModelProvider.OpenAI,
   responses: {
-    handlePayload: (payload: ChatStreamPayload) => {
-      const { model } = payload;
+    handlePayload: (payload) => {
+      const { enabledSearch, model, tools, ...rest } = payload;
+
+      const openaiTools = enabledSearch
+        ? [
+            ...(tools || []),
+            {
+              type: 'web_search_preview',
+              ...(oaiSearchContextSize && {
+                search_context_size: oaiSearchContextSize,
+              }),
+            },
+          ]
+        : tools;
+
       if (prunePrefixes.some((prefix) => model.startsWith(prefix))) {
         if (!payload.reasoning) {
           payload.reasoning = { summary: 'auto' };
         } else {
           payload.reasoning.summary = 'auto';
         }
+
+        // computer-use series must set truncation as auto
+        if (model.startsWith('computer-use')) {
+          payload.truncation = 'auto';
+        }
+
+        return pruneReasoningPayload(payload) as any;
       }
 
-      return { ...payload, stream: payload.stream ?? true };
+      return { ...rest, model, stream: payload.stream ?? true, tools: openaiTools } as any;
     },
   },
 });

package/src/libs/model-runtime/types/chat.ts

@@ -107,6 +107,7 @@ export interface ChatStreamPayload {
     effort?: string;
     summary?: string;
   };
+  reasoning_effort?: 'low' | 'medium' | 'high';
   responseMode?: 'stream' | 'json';
   /**
    * @title 是否开启流式请求
@@ -132,6 +133,7 @@ export interface ChatStreamPayload {
    * @default 1
    */
   top_p?: number;
+  truncation?: 'auto' | 'disabled';
 }
 
 export interface ChatMethodOptions {

package/src/libs/model-runtime/utils/openaiCompatibleFactory/index.ts

@@ -209,14 +209,9 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
     }
 
     async chat(
-      { responseMode, apiMode, ...payload }: ChatStreamPayload,
+      { responseMode, ...payload }: ChatStreamPayload,
       options?: ChatMethodOptions,
     ) {
-      // new openai Response API
-      if (apiMode === 'responses') {
-        return this.handleResponseAPIMode(payload, options);
-      }
-
       try {
         const inputStartAt = Date.now();
         const postPayload = chatCompletion?.handlePayload
@@ -226,6 +221,11 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
               stream: payload.stream ?? true,
             } as OpenAI.ChatCompletionCreateParamsStreaming);
 
+        // new openai Response API
+        if ((postPayload as any).apiMode === 'responses') {
+          return this.handleResponseAPIMode(payload, options);
+        }
+
         const messages = await convertOpenAIMessages(postPayload.messages);
 
         let response: Stream<OpenAI.Chat.Completions.ChatCompletionChunk>;
@@ -478,11 +478,12 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
     ): Promise<Response> {
       const inputStartAt = Date.now();
 
-      const { messages, ...res } = responses?.handlePayload
+      const { messages, reasoning_effort, tools, ...res } = responses?.handlePayload
         ? (responses?.handlePayload(payload, this._options) as ChatStreamPayload)
         : payload;
 
       // remove penalty params
+      delete res.apiMode;
       delete res.frequency_penalty;
       delete res.presence_penalty;
 
@@ -490,9 +491,10 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
 
       const postPayload = {
         ...res,
+        ...(reasoning_effort ? { reasoning: { effort: reasoning_effort } } : {}),
         input,
         store: false,
-        tools: payload.tools?.map((tool) => this.convertChatCompletionToolToResponseTool(tool)),
+        tools: tools?.map((tool) => this.convertChatCompletionToolToResponseTool(tool)),
       } as OpenAI.Responses.ResponseCreateParamsStreaming;
 
       if (debug?.responses?.()) {

package/src/libs/model-runtime/utils/streams/openai/__snapshots__/responsesStream.test.ts.snap

@@ -86,11 +86,11 @@ exports[`OpenAIResponsesStream > Reasoning > summary 1`] = `
   "data: " analyzing"
 
 ",
-  "id: resp_684313b89200819087f27686e0c822260b502bf083132d0d
+  "id: rs_684313b9774481908ee856625f82fb8c0b502bf083132d0d
 ",
-  "event: data
+  "event: text
 ",
-  "data: {"type":"response.output_item.done","output_index":0,"item":{"id":"rs_684313b9774481908ee856625f82fb8c0b502bf083132d0d","type":"reasoning","summary":[{"type":"summary_text","text":"**Answering a numeric comparison**\\n\\nThe user is asking in Chinese which number is larger: 9.1 or 9.92. This is straightforward since 9.92 is clearly larger, as it's greater than 9.1. We can respond with \\"9.92大于9.1\\" without needing to search for more information. It's simple comparison, but I could also add a little explanation, noting that 9.92 is indeed 0.82 more than 9.1. However, keeping it simple with \\"9.92 > 9.1\\" is perfectly fine!"}]}}
+  "data: null
 
 ",
   "id: resp_684313b89200819087f27686e0c822260b502bf083132d0d
@@ -128,11 +128,11 @@ exports[`OpenAIResponsesStream > Reasoning > summary 1`] = `
   "data: {"type":"response.content_part.done","item_id":"msg_684313bee2c88190b0f4b09621ad7dc60b502bf083132d0d","output_index":1,"content_index":0,"part":{"type":"output_text","annotations":[],"text":"9.92 比 9.1 大。"}}
 
 ",
-  "id: resp_684313b89200819087f27686e0c822260b502bf083132d0d
+  "id: msg_684313bee2c88190b0f4b09621ad7dc60b502bf083132d0d
 ",
-  "event: data
+  "event: text
 ",
-  "data: {"type":"response.output_item.done","output_index":1,"item":{"id":"msg_684313bee2c88190b0f4b09621ad7dc60b502bf083132d0d","type":"message","status":"completed","content":[{"type":"output_text","annotations":[],"text":"9.92 比 9. 大。"}],"role":"assistant"}}
+  "data: null
 
 ",
   "id: resp_684313b89200819087f27686e0c822260b502bf083132d0d

package/src/libs/model-runtime/utils/streams/openai/responsesStream.ts

@@ -1,7 +1,7 @@
 import OpenAI from 'openai';
 import type { Stream } from 'openai/streaming';
 
-import { ChatMessageError } from '@/types/message';
+import { ChatMessageError, CitationItem } from '@/types/message';
 
 import { AgentRuntimeErrorType } from '../../../error';
 import { convertResponseUsage } from '../../usageConverter';
@@ -20,7 +20,17 @@ import {
 import { OpenAIStreamOptions } from './openai';
 
 const transformOpenAIStream = (
-  chunk: OpenAI.Responses.ResponseStreamEvent,
+  chunk: OpenAI.Responses.ResponseStreamEvent | {
+    annotation: {
+      end_index: number;
+      start_index: number;
+      title: string;
+      type: 'url_citation';
+      url: string;
+    };
+    item_id: string;
+    type: 'response.output_text.annotation.added';
+  },
   streamContext: StreamContext,
 ): StreamProtocolChunk | StreamProtocolChunk[] => {
   // handle the first chunk error
@@ -42,6 +52,7 @@ const transformOpenAIStream = (
     switch (chunk.type) {
       case 'response.created': {
         streamContext.id = chunk.response.id;
+        streamContext.returnedCitationArray = [];
 
         return { data: chunk.response.status, id: streamContext.id, type: 'data' };
       }
@@ -106,6 +117,31 @@ const transformOpenAIStream = (
         return { data: chunk.delta, id: chunk.item_id, type: 'reasoning' };
       }
 
+      case 'response.output_text.annotation.added': {
+        const citations = chunk.annotation;
+
+        if (streamContext.returnedCitationArray) {
+          streamContext.returnedCitationArray.push({
+            title: citations.title,
+            url: citations.url,
+          } as CitationItem);
+        }
+
+        return { data: null, id: chunk.item_id, type: 'text' };
+      }
+
+      case 'response.output_item.done': {
+        if (streamContext.returnedCitationArray?.length) {
+          return {
+            data: { citations: streamContext.returnedCitationArray },
+            id: chunk.item.id,
+            type: 'grounding',
+          }
+        }
+
+        return { data: null, id: chunk.item.id, type: 'text' };
+      }
+
       case 'response.completed': {
         if (chunk.response.usage) {
           return {

package/src/services/chat.ts

@@ -388,6 +388,13 @@ class ChatService {
     const userPreferTransitionMode =
       userGeneralSettingsSelectors.transitionMode(getUserStoreState());
 
+    // The order of the array is very important.
+    const mergedResponseAnimation = [
+      providerConfig?.settings?.responseAnimation || {},
+      userPreferTransitionMode,
+      responseAnimation,
+    ].reduce((acc, cur) => merge(acc, standardizeAnimationStyle(cur)), {});
+
     return fetchSSE(API_ENDPOINTS.chat(sdkType), {
       body: JSON.stringify(payload),
       fetcher: fetcher,
@@ -397,10 +404,7 @@ class ChatService {
       onErrorHandle: options?.onErrorHandle,
       onFinish: options?.onFinish,
       onMessageHandle: options?.onMessageHandle,
-      responseAnimation: [userPreferTransitionMode, responseAnimation].reduce(
-        (acc, cur) => merge(acc, standardizeAnimationStyle(cur)),
-        providerConfig?.settings?.responseAnimation ?? {},
-      ),
+      responseAnimation: mergedResponseAnimation,
       signal,
     });
   };