Merge branch 'main' into feat/sora2

# Conflicts:
#	package.json
#	src/renderer/src/config/models/utils.ts
#	src/renderer/src/store/migrate.ts

.github/workflows/claude-translator.yml

@@ -16,10 +16,13 @@ on:
 jobs:
   translate:
     if: |
-      (github.event_name == 'issues') ||
-      (github.event_name == 'issue_comment' && github.event.sender.type != 'Bot') ||
-      (github.event_name == 'pull_request_review' && github.event.sender.type != 'Bot') ||
-      (github.event_name == 'pull_request_review_comment' && github.event.sender.type != 'Bot')
+      (github.event_name == 'issues')
+      || (github.event_name == 'issue_comment' && github.event.sender.type != 'Bot')
+      || (
+        (github.event_name == 'pull_request_review' || github.event_name == 'pull_request_review_comment')
+        && github.event.sender.type != 'Bot'
+        && github.event.pull_request.head.repo.fork == false
+      )
     runs-on: ubuntu-latest
     permissions:
       contents: read
@@ -42,7 +45,7 @@ jobs:
           # See: https://github.com/anthropics/claude-code-action/blob/main/docs/security.md
           github_token: ${{ secrets.TOKEN_GITHUB_WRITE }}
           allowed_non_write_users: "*"
-          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+          anthropic_api_key: ${{ secrets.CLAUDE_TRANSLATOR_APIKEY }}
           claude_args: "--allowed-tools Bash(gh issue:*),Bash(gh api:repos/*/issues:*),Bash(gh api:repos/*/pulls/*/reviews/*),Bash(gh api:repos/*/pulls/comments/*)"
           prompt: |
             你是一个多语言翻译助手。你需要响应 GitHub Webhooks 中的以下四种事件：
@@ -105,3 +108,5 @@ jobs:
 
             使用以下命令获取完整信息：
             gh issue view ${{ github.event.issue.number }} --json title,body,comments
+        env:
+          ANTHROPIC_BASE_URL: ${{ secrets.CLAUDE_TRANSLATOR_BASEURL }}

.yarn/patches/@ai-sdk-google-npm-2.0.20-b9102f9d54.patch

@@ -1,8 +1,8 @@
 diff --git a/dist/index.mjs b/dist/index.mjs
-index b957cb824faa79cf01ba3a504f221870bd8e306a..4d71d30f655775d61537d9d8b73f6e17d41fa67e 100644
+index 69ab1599c76801dc1167551b6fa283dded123466..f0af43bba7ad1196fe05338817e65b4ebda40955 100644
 --- a/dist/index.mjs
 +++ b/dist/index.mjs
-@@ -452,7 +452,7 @@ function convertToGoogleGenerativeAIMessages(prompt, options) {
+@@ -477,7 +477,7 @@ function convertToGoogleGenerativeAIMessages(prompt, options) {
  
  // src/get-model-path.ts
  function getModelPath(modelId) {

.yarn/patches/@opeoginni-github-copilot-openai-compatible-npm-0.1.18-3f65760532.patch

@@ -0,0 +1,44 @@
+diff --git a/dist/index.js b/dist/index.js
+index 53f411e55a4c9a06fd29bb4ab8161c4ad15980cd..71b91f196c8b886ed90dd237dec5625d79d5677e 100644
+--- a/dist/index.js
++++ b/dist/index.js
+@@ -12676,10 +12676,13 @@ var OpenAIResponsesLanguageModel = class {
+                   }
+                 });
+               } else if (value.item.type === "message") {
+-                controller.enqueue({
+-                  type: "text-end",
+-                  id: value.item.id
+-                });
++                // Fix for gpt-5-codex: use currentTextId to ensure text-end matches text-start
++                if (currentTextId) {
++                  controller.enqueue({
++                    type: "text-end",
++                    id: currentTextId
++                  });
++                }
+                 currentTextId = null;
+               } else if (isResponseOutputItemDoneReasoningChunk(value)) {
+                 const activeReasoningPart = activeReasoning[value.item.id];
+diff --git a/dist/index.mjs b/dist/index.mjs
+index 7719264da3c49a66c2626082f6ccaae6e3ef5e89..090fd8cf142674192a826148428ed6a0c4a54e35 100644
+--- a/dist/index.mjs
++++ b/dist/index.mjs
+@@ -12670,10 +12670,13 @@ var OpenAIResponsesLanguageModel = class {
+                   }
+                 });
+               } else if (value.item.type === "message") {
+-                controller.enqueue({
+-                  type: "text-end",
+-                  id: value.item.id
+-                });
++                // Fix for gpt-5-codex: use currentTextId to ensure text-end matches text-start
++                if (currentTextId) {
++                  controller.enqueue({
++                    type: "text-end",
++                    id: currentTextId
++                  });
++                }
+                 currentTextId = null;
+               } else if (isResponseOutputItemDoneReasoningChunk(value)) {
+                 const activeReasoningPart = activeReasoning[value.item.id];

electron-builder.yml

@@ -125,21 +125,113 @@ afterSign: scripts/notarize.js
 artifactBuildCompleted: scripts/artifact-build-completed.js
 releaseInfo:
   releaseNotes: |
-    What's New in v1.6.3
+    <!--LANG:en-->
+    What's New in v1.7.0-beta.1
 
-    Features:
-    - Notes: Add spell-check control, automatic table line wrapping, export functionality, and LLM-based renaming
-    - UI: Expand topic rename clickable area, add middle-click tab closing, remove redundant scrollbars, fix message menubar overflow
-    - Editor: Add read-only extension support, make TextFilePreview read-only but copyable
-    - Models: Update support for DeepSeek v3.2, Claude 4.5, GLM 4.6, Gemini regex, and vision models
-    - Code Tools: Add GitHub Copilot CLI integration
+    Major Features:
+    - Agent System: Introducing intelligent Agent capabilities alongside Assistants. Agents can autonomously solve complex problems using Claude Code SDK with tool calling, file operations, and multi-turn reasoning
+    - Agent Management: Create, configure, and manage agents with custom settings including model selection, tool permissions, accessible paths, and MCP server integrations
+    - Agent Sessions: Dedicated session management for agent interactions with persistent message history and context tracking
+    - Unified UI: Streamlined interface combining Assistants and Agents tabs with improved navigation and settings management
+
+    Agent Features:
+    - Tool Support: Web search, file operations, bash commands, and custom MCP tools
+    - Advanced Configuration: Max turns, temperature, token limits
+    - Permission Control: Configurable tool approval modes (manual, automatic, none)
+    - Session Persistence: Automatic message saving with optimized streaming and database integration
+    - Model Selection: API-based model filtering with provider-specific support
+
+    UI/UX Improvements:
+    - Unified assistant/agent tabs with smooth animations
+    - In-place session name editing
+    - Virtual list rendering for improved performance
+    - Session count indicators for active agents
+    - Enhanced settings popup with tabbed interface
+    - Webview keyboard shortcut interception for search functionality
+
+    API & Infrastructure:
+    - RESTful API for agent and session management
+    - Drizzle ORM integration for agent database
+    - OAuth support for Claude Code authentication
+    - Express validator for request validation
+    - Comprehensive error handling with Zod schemas
+
+    Model Updates:
+    - Gemini 2.5 Image Flash support
+    - Grok 4 Fast with reasoning capabilities
+    - Qwen3-omni and Qwen3-vl thinking models
+    - DeepSeek, Claude 4.5, GLM 4.6 support
+    - GitHub Copilot CLI integration with gpt-5-codex
 
     Bug Fixes:
+    - Fix Swagger UI accessibility issues
+    - Fix AI SDK error display with syntax highlighting
+    - Fix webview search shortcut handling
+    - Fix agent model visibility for CherryIn provider
+    - Fix session message ordering and persistence
+    - Fix anthropic model visibility in agent configuration
+    - Fix knowledge base deletion and web search RAG errors
     - Fix migration for missing providers
-    - Fix forked topic retaining old name after rename
-    - Restore first token latency reporting in metrics
-    - Fix UI scrollbar and overflow issues
 
     Technical Updates:
-    - Upgrade to Electron 37.6.0
-    - Update dependencies across packages
+    - React 19.2.0 upgrade
+    - Enhanced Claude Code service with streaming support
+    - Improved message transformation and streaming lifecycle
+    - Database migration system with automatic schema sync
+    - Optimized bundle size and dependency management
+
+    <!--LANG:zh-CN-->
+    v1.7.0-beta.1 新特性
+
+    核心功能:
+    - Agent 系统：引入智能 Agent 能力,与助手(Assistant)并存。Agent 基于 Claude Code SDK 构建,具备工具调用、文件操作和多轮推理能力,可自主解决复杂问题
+    - Agent 管理：创建、配置和管理 Agent,支持自定义模型选择、工具权限、可访问路径和 MCP 服务器集成
+    - Agent 会话：专属会话管理系统,支持持久化消息历史和上下文追踪
+    - 统一界面：精简的助手和 Agent 标签页界面,改进导航和设置管理体验
+
+    Agent 功能特性:
+    - 工具支持：网页搜索、文件操作、Bash 命令执行和自定义 MCP 工具
+    - 高级配置：最大轮次、温度、Token 限制
+    - 权限控制：可配置的工具批准模式(手动、自动、无需批准)
+    - 会话持久化：自动消息保存,优化的流式传输和数据库集成
+    - 模型选择：基于 API 的模型过滤,支持特定提供商
+
+    界面与交互优化:
+    - 统一的助手/Agent 标签页,带有流畅动画效果
+    - 会话名称原地编辑功能
+    - 虚拟列表渲染,提升性能表现
+    - 活跃 Agent 的会话计数指示器
+    - 增强的设置弹窗,采用标签页界面
+    - Webview 键盘快捷键拦截,支持搜索功能
+
+    API 与基础设施:
+    - RESTful API 用于 Agent 和会话管理
+    - 集成 Drizzle ORM 管理 Agent 数据库
+    - Claude Code OAuth 认证支持
+    - Express validator 请求验证
+    - 基于 Zod 模式的完善错误处理
+
+    模型更新:
+    - 支持 Gemini 2.5 Image Flash
+    - Grok 4 Fast 推理能力
+    - Qwen3-omni 和 Qwen3-vl 思考模型
+    - DeepSeek、Claude 4.5、GLM 4.6 支持
+    - GitHub Copilot CLI 集成 gpt-5-codex
+
+    问题修复:
+    - 修复 Swagger UI 无法打开
+    - 修复 AI SDK 错误显示,添加语法高亮
+    - 修复 Webview 搜索快捷键处理
+    - 修复 CherryIn 提供商的 Agent 模型可见性
+    - 修复会话消息排序和持久化
+    - 修复 Anthropic 模型在 Agent 配置中的可见性
+    - 修复知识库删除和网页搜索 RAG 错误
+    - 修复缺失提供商的迁移问题
+
+    技术更新:
+    - 升级至 React 19.2.0
+    - 增强 Claude Code 服务流式传输支持
+    - 改进消息转换和流式生命周期
+    - 数据库迁移系统,支持自动模式同步
+    - 优化打包大小和依赖管理
+    <!--LANG:END-->

eslint.config.mjs

@@ -2,6 +2,7 @@ import tseslint from '@electron-toolkit/eslint-config-ts'
 import eslint from '@eslint/js'
 import eslintReact from '@eslint-react/eslint-plugin'
 import { defineConfig } from 'eslint/config'
+import importZod from 'eslint-plugin-import-zod'
 import oxlint from 'eslint-plugin-oxlint'
 import reactHooks from 'eslint-plugin-react-hooks'
 import simpleImportSort from 'eslint-plugin-simple-import-sort'
@@ -15,7 +16,8 @@ export default defineConfig([
   {
     plugins: {
       'simple-import-sort': simpleImportSort,
-      'unused-imports': unusedImports
+      'unused-imports': unusedImports,
+      'import-zod': importZod
     },
     rules: {
       '@typescript-eslint/explicit-function-return-type': 'off',
@@ -25,6 +27,7 @@ export default defineConfig([
       'simple-import-sort/exports': 'error',
       'unused-imports/no-unused-imports': 'error',
       '@eslint-react/no-prop-types': 'error',
+      'import-zod/prefer-zod-namespace': 'error'
     }
   },
   // Configuration for ensuring compatibility with the original ESLint(8.x) rules

package.json

@@ -1,6 +1,6 @@
 {
   "name": "CherryStudio",
-  "version": "1.7.0-alpha.5",
+  "version": "1.7.0-sora.3",
   "private": true,
   "description": "A powerful AI assistant for producer.",
   "main": "./out/main/index.js",
@@ -83,6 +83,7 @@
     "@libsql/win32-x64-msvc": "^0.4.7",
     "@napi-rs/system-ocr": "patch:@napi-rs/system-ocr@npm%3A1.0.2#~/.yarn/patches/@napi-rs-system-ocr-npm-1.0.2-59e7a78e8b.patch",
     "@strongtz/win32-arm64-msvc": "^0.4.7",
+    "express": "^5.1.0",
     "font-list": "^2.0.0",
     "graceful-fs": "^4.2.11",
     "jsdom": "26.1.0",
@@ -92,6 +93,7 @@
     "selection-hook": "^1.0.12",
     "sharp": "^0.34.3",
     "swagger-jsdoc": "^6.2.8",
+    "swagger-ui-express": "^5.0.1",
     "tesseract.js": "patch:tesseract.js@npm%3A6.0.1#~/.yarn/patches/tesseract.js-npm-6.0.1-2562a7e46d.patch",
     "turndown": "7.2.0"
   },
@@ -124,6 +126,7 @@
     "@cherrystudio/embedjs-ollama": "^0.1.31",
     "@cherrystudio/embedjs-openai": "^0.1.31",
     "@cherrystudio/extension-table-plus": "workspace:^",
+    "@cherrystudio/openai": "6.3.0-fork.1",
     "@dnd-kit/core": "^6.3.1",
     "@dnd-kit/modifiers": "^9.0.0",
     "@dnd-kit/sortable": "^10.0.0",
@@ -152,7 +155,7 @@
     "@opentelemetry/sdk-trace-base": "^2.0.0",
     "@opentelemetry/sdk-trace-node": "^2.0.0",
     "@opentelemetry/sdk-trace-web": "^2.0.0",
-    "@opeoginni/github-copilot-openai-compatible": "0.1.18",
+    "@opeoginni/github-copilot-openai-compatible": "patch:@opeoginni/github-copilot-openai-compatible@npm%3A0.1.18#~/.yarn/patches/@opeoginni-github-copilot-openai-compatible-npm-0.1.18-3f65760532.patch",
     "@playwright/test": "^1.52.0",
     "@radix-ui/react-context-menu": "^2.2.16",
     "@reduxjs/toolkit": "^2.2.5",
@@ -257,11 +260,11 @@
     "emoji-picker-element": "^1.22.1",
     "epub": "patch:epub@npm%3A1.3.0#~/.yarn/patches/epub-npm-1.3.0-8325494ffe.patch",
     "eslint": "^9.22.0",
+    "eslint-plugin-import-zod": "^1.2.0",
     "eslint-plugin-oxlint": "^1.15.0",
     "eslint-plugin-react-hooks": "^5.2.0",
     "eslint-plugin-simple-import-sort": "^12.1.1",
     "eslint-plugin-unused-imports": "^4.1.4",
-    "express": "^5.1.0",
     "express-validator": "^7.2.1",
     "fast-diff": "^1.3.0",
     "fast-xml-parser": "^5.2.0",
@@ -294,16 +297,15 @@
     "motion": "^12.10.5",
     "notion-helper": "^1.3.22",
     "npx-scope-finder": "^1.2.0",
-    "openai": "patch:openai@npm%3A5.12.2#~/.yarn/patches/openai-npm-5.12.2-30b075401c.patch",
-    "oxlint": "^1.15.0",
+    "oxlint": "^1.22.0",
     "oxlint-tsgolint": "^0.2.0",
     "p-queue": "^8.1.0",
     "pdf-lib": "^1.17.1",
     "pdf-parse": "^1.1.1",
     "playwright": "^1.52.0",
     "proxy-agent": "^6.5.0",
-    "react": "^19.0.0",
-    "react-dom": "^19.0.0",
+    "react": "^19.2.0",
+    "react-dom": "^19.2.0",
     "react-error-boundary": "^6.0.0",
     "react-hotkeys-hook": "^4.6.1",
     "react-i18next": "^14.1.2",
@@ -335,7 +337,6 @@
     "string-width": "^7.2.0",
     "striptags": "^3.2.0",
     "styled-components": "^6.1.11",
-    "swagger-ui-express": "^5.0.1",
     "swr": "^2.3.6",
     "tailwindcss": "^4.1.13",
     "tar": "^7.4.3",
@@ -371,17 +372,19 @@
     "app-builder-lib@npm:26.0.13": "patch:app-builder-lib@npm%3A26.0.13#~/.yarn/patches/app-builder-lib-npm-26.0.13-a064c9e1d0.patch",
     "app-builder-lib@npm:26.0.15": "patch:app-builder-lib@npm%3A26.0.15#~/.yarn/patches/app-builder-lib-npm-26.0.15-360e5b0476.patch",
     "atomically@npm:^1.7.0": "patch:atomically@npm%3A1.7.0#~/.yarn/patches/atomically-npm-1.7.0-e742e5293b.patch",
+    "esbuild": "^0.25.0",
     "file-stream-rotator@npm:^0.6.1": "patch:file-stream-rotator@npm%3A0.6.1#~/.yarn/patches/file-stream-rotator-npm-0.6.1-eab45fb13d.patch",
     "libsql@npm:^0.4.4": "patch:libsql@npm%3A0.4.7#~/.yarn/patches/libsql-npm-0.4.7-444e260fb1.patch",
     "node-abi": "4.12.0",
-    "openai@npm:^4.77.0": "patch:openai@npm%3A5.12.2#~/.yarn/patches/openai-npm-5.12.2-30b075401c.patch",
-    "openai@npm:^4.87.3": "patch:openai@npm%3A5.12.2#~/.yarn/patches/openai-npm-5.12.2-30b075401c.patch",
+    "openai@npm:^4.77.0": "npm:@cherrystudio/openai@6.3.0-fork.1",
+    "openai@npm:^4.87.3": "npm:@cherrystudio/openai@6.3.0-fork.1",
     "pdf-parse@npm:1.1.1": "patch:pdf-parse@npm%3A1.1.1#~/.yarn/patches/pdf-parse-npm-1.1.1-04a6109b2a.patch",
     "pkce-challenge@npm:^4.1.0": "patch:pkce-challenge@npm%3A4.1.0#~/.yarn/patches/pkce-challenge-npm-4.1.0-fbc51695a3.patch",
+    "tar-fs": "^2.1.4",
     "undici": "6.21.2",
     "vite": "npm:rolldown-vite@latest",
     "tesseract.js@npm:*": "patch:tesseract.js@npm%3A6.0.1#~/.yarn/patches/tesseract.js-npm-6.0.1-2562a7e46d.patch",
-    "@ai-sdk/google@npm:2.0.17": "patch:@ai-sdk/google@npm%3A2.0.17#~/.yarn/patches/@ai-sdk-google-npm-2.0.17-fd88491de4.patch"
+    "@ai-sdk/google@npm:2.0.20": "patch:@ai-sdk/google@npm%3A2.0.20#~/.yarn/patches/@ai-sdk-google-npm-2.0.20-b9102f9d54.patch"
   },
   "packageManager": "yarn@4.9.1",
   "lint-staged": {

packages/aiCore/src/core/providers/schemas.ts

@@ -13,7 +13,7 @@ import { LanguageModelV2 } from '@ai-sdk/provider'
 import { createXai } from '@ai-sdk/xai'
 import { createOpenRouter } from '@openrouter/ai-sdk-provider'
 import { customProvider, Provider } from 'ai'
-import { z } from 'zod'
+import * as z from 'zod'
 
 /**
  * 基础 Provider IDs

packages/shared/IpcChannel.ts

@@ -53,6 +53,7 @@ export enum IpcChannel {
 
   Webview_SetOpenLinkExternal = 'webview:set-open-link-external',
   Webview_SetSpellCheckEnabled = 'webview:set-spell-check-enabled',
+  Webview_SearchHotkey = 'webview:search-hotkey',
 
   // Open
   Open_Path = 'open:path',
@@ -316,6 +317,7 @@ export enum IpcChannel {
   ApiServer_Stop = 'api-server:stop',
   ApiServer_Restart = 'api-server:restart',
   ApiServer_GetStatus = 'api-server:get-status',
+  // NOTE: This api is not be used.
   ApiServer_GetConfig = 'api-server:get-config',
 
   // Anthropic OAuth
@@ -335,6 +337,7 @@ export enum IpcChannel {
 
   // OCR
   OCR_ocr = 'ocr:ocr',
+  OCR_ListProviders = 'ocr:list-providers',
 
   // OVMS
   Ovms_AddModel = 'ovms:add-model',

packages/shared/config/types.ts

@@ -22,3 +22,12 @@ export type MCPProgressEvent = {
   callId: string
   progress: number // 0-1 range
 }
+
+export type WebviewKeyEvent = {
+  webviewId: number
+  key: string
+  control: boolean
+  meta: boolean
+  shift: boolean
+  alt: boolean
+}

resources/scripts/install-ovms.js

@@ -5,105 +5,171 @@ const { execSync } = require('child_process')
 const { downloadWithPowerShell } = require('./download')
 
 // Base URL for downloading OVMS binaries
-const OVMS_PKG_NAME = 'ovms250911.zip'
-const OVMS_RELEASE_BASE_URL = [`https://gitcode.com/gcw_ggDjjkY3/kjfile/releases/download/download/${OVMS_PKG_NAME}`]
+const OVMS_RELEASE_BASE_URL =
+  'https://storage.openvinotoolkit.org/repositories/openvino_model_server/packages/2025.3.0/ovms_windows_python_on.zip'
+const OVMS_EX_URL = 'https://gitcode.com/gcw_ggDjjkY3/kjfile/releases/download/download/ovms_25.3_ex.zip'
 
 /**
- * Downloads and extracts the OVMS binary for the specified platform
+ * error code:
+ * 101: Unsupported CPU (not Intel Ultra)
+ * 102: Unsupported platform (not Windows)
+ * 103: Download failed
+ * 104: Installation failed
+ * 105: Failed to create ovdnd.exe
+ * 106: Failed to create run.bat
+ * 110: Cleanup of old installation failed
  */
-async function downloadOvmsBinary() {
-  // Create output directory structure - OVMS goes into its own subdirectory
+
+/**
+ * Clean old OVMS installation if it exists
+ */
+function cleanOldOvmsInstallation() {
+  console.log('Cleaning the existing OVMS installation...')
   const csDir = path.join(os.homedir(), '.cherrystudio')
-
-  // Ensure directories exist
-  fs.mkdirSync(csDir, { recursive: true })
-
   const csOvmsDir = path.join(csDir, 'ovms')
-  // Delete existing OVMS directory if it exists
   if (fs.existsSync(csOvmsDir)) {
-    fs.rmSync(csOvmsDir, { recursive: true })
-  }
-
-  const tempdir = os.tmpdir()
-  const tempFilename = path.join(tempdir, 'ovms.zip')
-
-  // Try each URL until one succeeds
-  let downloadSuccess = false
-  let lastError = null
-
-  for (let i = 0; i < OVMS_RELEASE_BASE_URL.length; i++) {
-    const downloadUrl = OVMS_RELEASE_BASE_URL[i]
-    console.log(`Attempting download from URL ${i + 1}/${OVMS_RELEASE_BASE_URL.length}: ${downloadUrl}`)
-
     try {
-      console.log(`Downloading OVMS from ${downloadUrl} to ${tempFilename}...`)
-
-      // Try PowerShell download first, fallback to Node.js download if it fails
-      await downloadWithPowerShell(downloadUrl, tempFilename)
-
-      // If we get here, download was successful
-      downloadSuccess = true
-      console.log(`Successfully downloaded from: ${downloadUrl}`)
-      break
+      fs.rmSync(csOvmsDir, { recursive: true })
     } catch (error) {
-      console.warn(`Download failed from ${downloadUrl}: ${error.message}`)
-      lastError = error
-
-      // Clean up failed download file if it exists
-      if (fs.existsSync(tempFilename)) {
-        try {
-          fs.unlinkSync(tempFilename)
-        } catch (cleanupError) {
-          console.warn(`Failed to clean up temporary file: ${cleanupError.message}`)
-        }
-      }
-
-      // Continue to next URL if this one failed
-      if (i < OVMS_RELEASE_BASE_URL.length - 1) {
-        console.log(`Trying next URL...`)
-      }
+      console.warn(`Failed to clean up old OVMS installation: ${error.message}`)
+      return 110
     }
   }
 
-  // Check if any download succeeded
-  if (!downloadSuccess) {
-    console.error(`All download URLs failed. Last error: ${lastError?.message || 'Unknown error'}`)
+  return 0
+}
+
+/**
+ * Install OVMS Base package
+ */
+async function installOvmsBase() {
+  // Download the base package
+  const tempdir = os.tmpdir()
+  const tempFilename = path.join(tempdir, 'ovms.zip')
+
+  try {
+    console.log(`Downloading OVMS Base Package from ${OVMS_RELEASE_BASE_URL} to ${tempFilename}...`)
+
+    // Try PowerShell download first, fallback to Node.js download if it fails
+    await downloadWithPowerShell(OVMS_RELEASE_BASE_URL, tempFilename)
+    console.log(`Successfully downloaded from: ${OVMS_RELEASE_BASE_URL}`)
+  } catch (error) {
+    console.error(`Download OVMS Base failed: ${error.message}`)
+    fs.unlinkSync(tempFilename)
     return 103
   }
 
-  try {
-    console.log(`Extracting to ${csDir}...`)
+  // unzip the base package to the target directory
+  const csDir = path.join(os.homedir(), '.cherrystudio')
+  const csOvmsDir = path.join(csDir, 'ovms')
+  fs.mkdirSync(csOvmsDir, { recursive: true })
 
+  try {
     // Use tar.exe to extract the ZIP file
-    console.log(`Extracting OVMS to ${csDir}...`)
-    execSync(`tar -xf ${tempFilename} -C ${csDir}`, { stdio: 'inherit' })
-    console.log(`OVMS extracted to ${csDir}`)
+    console.log(`Extracting OVMS Base to ${csOvmsDir}...`)
+    execSync(`tar -xf ${tempFilename} -C ${csOvmsDir}`, { stdio: 'inherit' })
+    console.log(`OVMS extracted to ${csOvmsDir}`)
 
     // Clean up temporary file
     fs.unlinkSync(tempFilename)
     console.log(`Installation directory: ${csDir}`)
   } catch (error) {
     console.error(`Error installing OVMS: ${error.message}`)
-    if (fs.existsSync(tempFilename)) {
-      fs.unlinkSync(tempFilename)
-    }
-
-    // Check if ovmsDir is empty and remove it if so
-    try {
-      const ovmsDir = path.join(csDir, 'ovms')
-      const files = fs.readdirSync(ovmsDir)
-      if (files.length === 0) {
-        fs.rmSync(ovmsDir, { recursive: true })
-        console.log(`Removed empty directory: ${ovmsDir}`)
-      }
-    } catch (cleanupError) {
-      console.warn(`Warning: Failed to clean up directory: ${cleanupError.message}`)
-      return 105
-    }
-
+    fs.unlinkSync(tempFilename)
     return 104
   }
 
+  const csOvmsBinDir = path.join(csOvmsDir, 'ovms')
+  // copy ovms.exe to ovdnd.exe
+  try {
+    fs.copyFileSync(path.join(csOvmsBinDir, 'ovms.exe'), path.join(csOvmsBinDir, 'ovdnd.exe'))
+    console.log('Copied ovms.exe to ovdnd.exe')
+  } catch (error) {
+    console.error(`Error copying ovms.exe to ovdnd.exe: ${error.message}`)
+    return 105
+  }
+
+  // copy {csOvmsBinDir}/setupvars.bat to {csOvmsBinDir}/run.bat, and append the following lines to run.bat:
+  // del %USERPROFILE%\.cherrystudio\ovms_log.log
+  // ovms.exe --config_path models/config.json --rest_port 8000 --log_level DEBUG --log_path %USERPROFILE%\.cherrystudio\ovms_log.log
+  const runBatPath = path.join(csOvmsBinDir, 'run.bat')
+  try {
+    fs.copyFileSync(path.join(csOvmsBinDir, 'setupvars.bat'), runBatPath)
+    fs.appendFileSync(runBatPath, '\r\n')
+    fs.appendFileSync(runBatPath, 'del %USERPROFILE%\\.cherrystudio\\ovms_log.log\r\n')
+    fs.appendFileSync(
+      runBatPath,
+      'ovms.exe --config_path models/config.json --rest_port 8000 --log_level DEBUG --log_path %USERPROFILE%\\.cherrystudio\\ovms_log.log\r\n'
+    )
+    console.log(`Created run.bat at: ${runBatPath}`)
+  } catch (error) {
+    console.error(`Error creating run.bat: ${error.message}`)
+    return 106
+  }
+
+  // create {csOvmsBinDir}/models/config.json with content '{"model_config_list": []}'
+  const configJsonPath = path.join(csOvmsBinDir, 'models', 'config.json')
+  fs.mkdirSync(path.dirname(configJsonPath), { recursive: true })
+  fs.writeFileSync(configJsonPath, '{"mediapipe_config_list":[],"model_config_list":[]}')
+  console.log(`Created config file: ${configJsonPath}`)
+
+  return 0
+}
+
+/**
+ * Install OVMS Extra package
+ */
+async function installOvmsExtra() {
+  // Download the extra package
+  const tempdir = os.tmpdir()
+  const tempFilename = path.join(tempdir, 'ovms_ex.zip')
+
+  try {
+    console.log(`Downloading OVMS Extra Package from ${OVMS_EX_URL} to ${tempFilename}...`)
+
+    // Try PowerShell download first, fallback to Node.js download if it fails
+    await downloadWithPowerShell(OVMS_EX_URL, tempFilename)
+    console.log(`Successfully downloaded from: ${OVMS_EX_URL}`)
+  } catch (error) {
+    console.error(`Download OVMS Extra failed: ${error.message}`)
+    fs.unlinkSync(tempFilename)
+    return 103
+  }
+
+  // unzip the extra package to the target directory
+  const csDir = path.join(os.homedir(), '.cherrystudio')
+  const csOvmsDir = path.join(csDir, 'ovms')
+
+  try {
+    // Use tar.exe to extract the ZIP file
+    console.log(`Extracting OVMS Extra to ${csOvmsDir}...`)
+    execSync(`tar -xf ${tempFilename} -C ${csOvmsDir}`, { stdio: 'inherit' })
+    console.log(`OVMS extracted to ${csOvmsDir}`)
+
+    // Clean up temporary file
+    fs.unlinkSync(tempFilename)
+    console.log(`Installation directory: ${csDir}`)
+  } catch (error) {
+    console.error(`Error installing OVMS Extra: ${error.message}`)
+    fs.unlinkSync(tempFilename)
+    return 104
+  }
+
+  // apply ovms patch, copy all files in {csOvmsDir}/patch/ovms to {csOvmsDir}/ovms with overwrite mode
+  const patchDir = path.join(csOvmsDir, 'patch', 'ovms')
+  const csOvmsBinDir = path.join(csOvmsDir, 'ovms')
+  try {
+    const files = fs.readdirSync(patchDir)
+    files.forEach((file) => {
+      const srcPath = path.join(patchDir, file)
+      const destPath = path.join(csOvmsBinDir, file)
+      fs.copyFileSync(srcPath, destPath)
+      console.log(`Applied patch file: ${file}`)
+    })
+  } catch (error) {
+    console.error(`Error applying OVMS patch: ${error.message}`)
+  }
+
   return 0
 }
 
@@ -158,7 +224,27 @@ async function installOvms() {
     return 102
   }
 
-  return await downloadOvmsBinary()
+  // Clean old installation if it exists
+  const cleanupCode = cleanOldOvmsInstallation()
+  if (cleanupCode !== 0) {
+    console.error(`OVMS cleanup failed with code: ${cleanupCode}`)
+    return cleanupCode
+  }
+
+  const installBaseCode = await installOvmsBase()
+  if (installBaseCode !== 0) {
+    console.error(`OVMS Base installation failed with code: ${installBaseCode}`)
+    cleanOldOvmsInstallation()
+    return installBaseCode
+  }
+
+  const installExtraCode = await installOvmsExtra()
+  if (installExtraCode !== 0) {
+    console.error(`OVMS Extra installation failed with code: ${installExtraCode}`)
+    return installExtraCode
+  }
+
+  return 0
 }
 
 // Run the installation

scripts/auto-translate-i18n.ts

@@ -2,9 +2,9 @@
  * 该脚本用于少量自动翻译所有baseLocale以外的文本。待翻译文案必须以[to be translated]开头
  *
  */
+import OpenAI from '@cherrystudio/openai'
 import cliProgress from 'cli-progress'
 import * as fs from 'fs'
-import OpenAI from 'openai'
 import * as path from 'path'
 
 const localesDir = path.join(__dirname, '../src/renderer/src/i18n/locales')

scripts/update-i18n.ts

@@ -4,9 +4,9 @@
  * API_KEY=sk-xxxx BASE_URL=xxxx MODEL=xxxx ts-node scripts/update-i18n.ts
  */
 
+import OpenAI from '@cherrystudio/openai'
 import cliProgress from 'cli-progress'
 import fs from 'fs'
-import OpenAI from 'openai'
 
 type I18NValue = string | { [key: string]: I18NValue }
 type I18N = { [key: string]: I18NValue }

src/main/apiServer/routes/chat.ts

@@ -1,5 +1,5 @@
+import { ChatCompletionCreateParams } from '@cherrystudio/openai/resources'
 import express, { Request, Response } from 'express'
-import { ChatCompletionCreateParams } from 'openai/resources'
 
 import { loggerService } from '../../services/LoggerService'
 import {

src/main/apiServer/services/chat-completion.ts

@@ -1,6 +1,6 @@
+import OpenAI from '@cherrystudio/openai'
+import { ChatCompletionCreateParams, ChatCompletionCreateParamsStreaming } from '@cherrystudio/openai/resources'
 import { Provider } from '@types'
-import OpenAI from 'openai'
-import { ChatCompletionCreateParams, ChatCompletionCreateParamsStreaming } from 'openai/resources'
 
 import { loggerService } from '../../services/LoggerService'
 import { ModelValidationError, validateModelId } from '../utils'

src/main/apiServer/services/models.ts

@@ -1,6 +1,13 @@
+import { isEmpty } from 'lodash'
+
 import { ApiModel, ApiModelsFilter, ApiModelsResponse } from '../../../renderer/src/types/apiModels'
 import { loggerService } from '../../services/LoggerService'
-import { getAvailableProviders, listAllAvailableModels, transformModelToOpenAI } from '../utils'
+import {
+  getAvailableProviders,
+  getProviderAnthropicModelChecker,
+  listAllAvailableModels,
+  transformModelToOpenAI
+} from '../utils'
 
 const logger = loggerService.withContext('ModelsService')
 
@@ -16,9 +23,7 @@ export class ModelsService {
       let providers = await getAvailableProviders()
 
       if (filter.providerType === 'anthropic') {
-        providers = providers.filter(
-          (p) => p.type === 'anthropic' || (p.anthropicApiHost !== undefined && p.anthropicApiHost.trim() !== '')
-        )
+        providers = providers.filter((p) => p.type === 'anthropic' || !isEmpty(p.anthropicApiHost?.trim()))
       }
 
       const models = await listAllAvailableModels(providers)
@@ -27,18 +32,18 @@ export class ModelsService {
 
       for (const model of models) {
         const provider = providers.find((p) => p.id === model.provider)
-        logger.debug(`Processing model ${model.id} from provider ${model.provider}`, {
-          isAnthropicModel: provider?.isAnthropicModel
-        })
-        if (
-          !provider ||
-          (filter.providerType === 'anthropic' && provider.isAnthropicModel && !provider.isAnthropicModel(model))
-        ) {
+        logger.debug(`Processing model ${model.id}`)
+        if (!provider) {
+          logger.debug(`Skipping model ${model.id} . Reason: Provider not found.`)
           continue
         }
-        // Special case: For "aihubmix", it should be covered by above condition, but just in case
-        if (provider.id === 'aihubmix' && filter.providerType === 'anthropic' && !model.id.includes('claude')) {
-          continue
+
+        if (filter.providerType === 'anthropic') {
+          const checker = getProviderAnthropicModelChecker(provider.id)
+          if (!checker(model)) {
+            logger.debug(`Skipping model ${model.id} from ${model.provider}. Reason: Not an Anthropic model.`)
+            continue
+          }
         }
 
         const openAIModel = transformModelToOpenAI(model, provider)

src/main/apiServer/utils/index.ts

@@ -279,3 +279,16 @@ export function validateProvider(provider: Provider): boolean {
     return false
   }
 }
+
+export const getProviderAnthropicModelChecker = (providerId: string): ((m: Model) => boolean) => {
+  switch (providerId) {
+    case 'cherryin':
+    case 'new-api':
+      return (m: Model) => m.endpoint_type === 'anthropic'
+    case 'aihubmix':
+      return (m: Model) => m.id.includes('claude')
+    default:
+      // allow all models when checker not configured
+      return () => true
+  }
+}

src/main/index.ts

@@ -30,6 +30,7 @@ import selectionService, { initSelectionService } from './services/SelectionServ
 import { registerShortcuts } from './services/ShortcutService'
 import { TrayService } from './services/TrayService'
 import { windowService } from './services/WindowService'
+import { initWebviewHotkeys } from './services/WebviewService'
 
 const logger = loggerService.withContext('MainEntry')
 
@@ -108,6 +109,7 @@ if (!app.requestSingleInstanceLock()) {
   // Some APIs can only be used after this event occurs.
 
   app.whenReady().then(async () => {
+    initWebviewHotkeys()
     // Set app user model id for windows
     electronApp.setAppUserModelId(import.meta.env.VITE_MAIN_BUNDLE_ID || 'com.kangfenmao.CherryStudio')
 
@@ -157,11 +159,26 @@ if (!app.requestSingleInstanceLock()) {
       logger.error('Failed to initialize Agent service:', error)
     }
 
-    // Start API server if enabled
+    // Start API server if enabled or if agents exist
     try {
       const config = await apiServerService.getCurrentConfig()
       logger.info('API server config:', config)
-      if (config.enabled) {
+
+      // Check if there are any agents
+      let shouldStart = config.enabled
+      if (!shouldStart) {
+        try {
+          const { total } = await agentService.listAgents({ limit: 1 })
+          if (total > 0) {
+            shouldStart = true
+            logger.info(`Detected ${total} agent(s), auto-starting API server`)
+          }
+        } catch (error: any) {
+          logger.warn('Failed to check agent count:', error)
+        }
+      }
+
+      if (shouldStart) {
         await apiServerService.start()
       }
     } catch (error: any) {

src/main/ipc.ts

@@ -786,7 +786,6 @@ export function registerIpc(mainWindow: BrowserWindow, app: Electron.App) {
   ipcMain.handle(IpcChannel.Webview_SetOpenLinkExternal, (_, webviewId: number, isExternal: boolean) =>
     setOpenLinkExternal(webviewId, isExternal)
   )
-
   ipcMain.handle(IpcChannel.Webview_SetSpellCheckEnabled, (_, webviewId: number, isEnable: boolean) => {
     const webview = webContents.fromId(webviewId)
     if (!webview) return
@@ -876,6 +875,7 @@ export function registerIpc(mainWindow: BrowserWindow, app: Electron.App) {
   ipcMain.handle(IpcChannel.OCR_ocr, (_, file: SupportedOcrFile, provider: OcrProvider) =>
     ocrService.ocr(file, provider)
   )
+  ipcMain.handle(IpcChannel.OCR_ListProviders, () => ocrService.listProviderIds())
 
   // OVMS
   ipcMain.handle(IpcChannel.Ovms_AddModel, (_, modelName: string, modelId: string, modelSource: string, task: string) =>

src/main/mcpServers/didi-mcp.ts

@@ -0,0 +1,473 @@
+/**
+ * DiDi MCP Server Implementation
+ *
+ * Based on official DiDi MCP API capabilities.
+ * API Documentation: https://mcp.didichuxing.com/api?tap=api
+ *
+ * Provides ride-hailing services including map search, price estimation,
+ * order management, and driver tracking.
+ *
+ * Note: Only available in Mainland China.
+ */
+
+import { loggerService } from '@logger'
+import { Server } from '@modelcontextprotocol/sdk/server/index.js'
+import { CallToolRequestSchema, ListToolsRequestSchema } from '@modelcontextprotocol/sdk/types.js'
+
+const logger = loggerService.withContext('DiDiMCPServer')
+
+export class DiDiMcpServer {
+  private _server: Server
+  private readonly baseUrl = 'http://mcp.didichuxing.com/mcp-servers'
+  private apiKey: string
+
+  constructor(apiKey?: string) {
+    this._server = new Server(
+      {
+        name: 'didi-mcp-server',
+        version: '0.1.0'
+      },
+      {
+        capabilities: {
+          tools: {}
+        }
+      }
+    )
+
+    // Get API key from parameter or environment variables
+    this.apiKey = apiKey || process.env.DIDI_API_KEY || ''
+    if (!this.apiKey) {
+      logger.warn('DIDI_API_KEY environment variable is not set')
+    }
+
+    this.setupRequestHandlers()
+  }
+
+  get server(): Server {
+    return this._server
+  }
+
+  private setupRequestHandlers() {
+    // List available tools
+    this._server.setRequestHandler(ListToolsRequestSchema, async () => {
+      return {
+        tools: [
+          {
+            name: 'maps_textsearch',
+            description: 'Search for POI locations based on keywords and city',
+            inputSchema: {
+              type: 'object',
+              properties: {
+                city: {
+                  type: 'string',
+                  description: 'Query city'
+                },
+                keywords: {
+                  type: 'string',
+                  description: 'Search keywords'
+                },
+                location: {
+                  type: 'string',
+                  description: 'Location coordinates, format: longitude,latitude'
+                }
+              },
+              required: ['keywords', 'city']
+            }
+          },
+          {
+            name: 'taxi_cancel_order',
+            description: 'Cancel a taxi order',
+            inputSchema: {
+              type: 'object',
+              properties: {
+                order_id: {
+                  type: 'string',
+                  description: 'Order ID from order creation or query results'
+                },
+                reason: {
+                  type: 'string',
+                  description:
+                    'Cancellation reason (optional). Examples: no longer needed, waiting too long, urgent matter'
+                }
+              },
+              required: ['order_id']
+            }
+          },
+          {
+            name: 'taxi_create_order',
+            description: 'Create taxi order directly via API without opening any app interface',
+            inputSchema: {
+              type: 'object',
+              properties: {
+                caller_car_phone: {
+                  type: 'string',
+                  description: 'Caller phone number (optional)'
+                },
+                estimate_trace_id: {
+                  type: 'string',
+                  description: 'Estimation trace ID from estimation results'
+                },
+                product_category: {
+                  type: 'string',
+                  description: 'Vehicle category ID from estimation results, comma-separated for multiple types'
+                }
+              },
+              required: ['product_category', 'estimate_trace_id']
+            }
+          },
+          {
+            name: 'taxi_estimate',
+            description: 'Get available ride-hailing vehicle types and fare estimates',
+            inputSchema: {
+              type: 'object',
+              properties: {
+                from_lat: {
+                  type: 'string',
+                  description: 'Departure latitude, must be from map tools'
+                },
+                from_lng: {
+                  type: 'string',
+                  description: 'Departure longitude, must be from map tools'
+                },
+                from_name: {
+                  type: 'string',
+                  description: 'Departure location name'
+                },
+                to_lat: {
+                  type: 'string',
+                  description: 'Destination latitude, must be from map tools'
+                },
+                to_lng: {
+                  type: 'string',
+                  description: 'Destination longitude, must be from map tools'
+                },
+                to_name: {
+                  type: 'string',
+                  description: 'Destination name'
+                }
+              },
+              required: ['from_lng', 'from_lat', 'from_name', 'to_lng', 'to_lat', 'to_name']
+            }
+          },
+          {
+            name: 'taxi_generate_ride_app_link',
+            description: 'Generate deep links to open ride-hailing apps based on origin, destination and vehicle type',
+            inputSchema: {
+              type: 'object',
+              properties: {
+                from_lat: {
+                  type: 'string',
+                  description: 'Departure latitude, must be from map tools'
+                },
+                from_lng: {
+                  type: 'string',
+                  description: 'Departure longitude, must be from map tools'
+                },
+                product_category: {
+                  type: 'string',
+                  description: 'Vehicle category IDs from estimation results, comma-separated for multiple types'
+                },
+                to_lat: {
+                  type: 'string',
+                  description: 'Destination latitude, must be from map tools'
+                },
+                to_lng: {
+                  type: 'string',
+                  description: 'Destination longitude, must be from map tools'
+                }
+              },
+              required: ['from_lng', 'from_lat', 'to_lng', 'to_lat']
+            }
+          },
+          {
+            name: 'taxi_get_driver_location',
+            description: 'Get real-time driver location for a taxi order',
+            inputSchema: {
+              type: 'object',
+              properties: {
+                order_id: {
+                  type: 'string',
+                  description: 'Taxi order ID'
+                }
+              },
+              required: ['order_id']
+            }
+          },
+          {
+            name: 'taxi_query_order',
+            description: 'Query taxi order status and information such as driver contact, license plate, ETA',
+            inputSchema: {
+              type: 'object',
+              properties: {
+                order_id: {
+                  type: 'string',
+                  description: 'Order ID from order creation results, if available; otherwise queries incomplete orders'
+                }
+              }
+            }
+          }
+        ]
+      }
+    })
+
+    // Handle tool calls
+    this._server.setRequestHandler(CallToolRequestSchema, async (request) => {
+      const { name, arguments: args } = request.params
+
+      try {
+        switch (name) {
+          case 'maps_textsearch':
+            return await this.handleMapsTextSearch(args)
+          case 'taxi_cancel_order':
+            return await this.handleTaxiCancelOrder(args)
+          case 'taxi_create_order':
+            return await this.handleTaxiCreateOrder(args)
+          case 'taxi_estimate':
+            return await this.handleTaxiEstimate(args)
+          case 'taxi_generate_ride_app_link':
+            return await this.handleTaxiGenerateRideAppLink(args)
+          case 'taxi_get_driver_location':
+            return await this.handleTaxiGetDriverLocation(args)
+          case 'taxi_query_order':
+            return await this.handleTaxiQueryOrder(args)
+          default:
+            throw new Error(`Unknown tool: ${name}`)
+        }
+      } catch (error) {
+        logger.error(`Error calling tool ${name}:`, error as Error)
+        throw error
+      }
+    })
+  }
+
+  private async handleMapsTextSearch(args: any) {
+    const { city, keywords, location } = args
+
+    const params = {
+      name: 'maps_textsearch',
+      arguments: {
+        keywords,
+        city,
+        ...(location && { location })
+      }
+    }
+
+    try {
+      const response = await this.makeRequest('tools/call', params)
+      return {
+        content: [
+          {
+            type: 'text',
+            text: JSON.stringify(response, null, 2)
+          }
+        ]
+      }
+    } catch (error) {
+      logger.error('Maps text search error:', error as Error)
+      throw error
+    }
+  }
+
+  private async handleTaxiCancelOrder(args: any) {
+    const { order_id, reason } = args
+
+    const params = {
+      name: 'taxi_cancel_order',
+      arguments: {
+        order_id,
+        ...(reason && { reason })
+      }
+    }
+
+    try {
+      const response = await this.makeRequest('tools/call', params)
+      return {
+        content: [
+          {
+            type: 'text',
+            text: JSON.stringify(response, null, 2)
+          }
+        ]
+      }
+    } catch (error) {
+      logger.error('Taxi cancel order error:', error as Error)
+      throw error
+    }
+  }
+
+  private async handleTaxiCreateOrder(args: any) {
+    const { caller_car_phone, estimate_trace_id, product_category } = args
+
+    const params = {
+      name: 'taxi_create_order',
+      arguments: {
+        product_category,
+        estimate_trace_id,
+        ...(caller_car_phone && { caller_car_phone })
+      }
+    }
+
+    try {
+      const response = await this.makeRequest('tools/call', params)
+      return {
+        content: [
+          {
+            type: 'text',
+            text: JSON.stringify(response, null, 2)
+          }
+        ]
+      }
+    } catch (error) {
+      logger.error('Taxi create order error:', error as Error)
+      throw error
+    }
+  }
+
+  private async handleTaxiEstimate(args: any) {
+    const { from_lng, from_lat, from_name, to_lng, to_lat, to_name } = args
+
+    const params = {
+      name: 'taxi_estimate',
+      arguments: {
+        from_lng,
+        from_lat,
+        from_name,
+        to_lng,
+        to_lat,
+        to_name
+      }
+    }
+
+    try {
+      const response = await this.makeRequest('tools/call', params)
+      return {
+        content: [
+          {
+            type: 'text',
+            text: JSON.stringify(response, null, 2)
+          }
+        ]
+      }
+    } catch (error) {
+      logger.error('Taxi estimate error:', error as Error)
+      throw error
+    }
+  }
+
+  private async handleTaxiGenerateRideAppLink(args: any) {
+    const { from_lng, from_lat, to_lng, to_lat, product_category } = args
+
+    const params = {
+      name: 'taxi_generate_ride_app_link',
+      arguments: {
+        from_lng,
+        from_lat,
+        to_lng,
+        to_lat,
+        ...(product_category && { product_category })
+      }
+    }
+
+    try {
+      const response = await this.makeRequest('tools/call', params)
+      return {
+        content: [
+          {
+            type: 'text',
+            text: JSON.stringify(response, null, 2)
+          }
+        ]
+      }
+    } catch (error) {
+      logger.error('Taxi generate ride app link error:', error as Error)
+      throw error
+    }
+  }
+
+  private async handleTaxiGetDriverLocation(args: any) {
+    const { order_id } = args
+
+    const params = {
+      name: 'taxi_get_driver_location',
+      arguments: {
+        order_id
+      }
+    }
+
+    try {
+      const response = await this.makeRequest('tools/call', params)
+      return {
+        content: [
+          {
+            type: 'text',
+            text: JSON.stringify(response, null, 2)
+          }
+        ]
+      }
+    } catch (error) {
+      logger.error('Taxi get driver location error:', error as Error)
+      throw error
+    }
+  }
+
+  private async handleTaxiQueryOrder(args: any) {
+    const { order_id } = args
+
+    const params = {
+      name: 'taxi_query_order',
+      arguments: {
+        ...(order_id && { order_id })
+      }
+    }
+
+    try {
+      const response = await this.makeRequest('tools/call', params)
+      return {
+        content: [
+          {
+            type: 'text',
+            text: JSON.stringify(response, null, 2)
+          }
+        ]
+      }
+    } catch (error) {
+      logger.error('Taxi query order error:', error as Error)
+      throw error
+    }
+  }
+
+  private async makeRequest(method: string, params: any): Promise<any> {
+    const requestData = {
+      jsonrpc: '2.0',
+      method: method,
+      id: Date.now(),
+      ...(Object.keys(params).length > 0 && { params })
+    }
+
+    // API key is passed as URL parameter
+    const url = `${this.baseUrl}?key=${this.apiKey}`
+
+    const response = await fetch(url, {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json'
+      },
+      body: JSON.stringify(requestData)
+    })
+
+    if (!response.ok) {
+      const errorText = await response.text()
+      throw new Error(`HTTP ${response.status}: ${errorText}`)
+    }
+
+    const data = await response.json()
+
+    if (data.error) {
+      throw new Error(`API Error: ${JSON.stringify(data.error)}`)
+    }
+
+    return data.result
+  }
+}
+
+export default DiDiMcpServer

src/main/mcpServers/dify-knowledge.ts

@@ -3,7 +3,7 @@ import { loggerService } from '@logger'
 import { Server } from '@modelcontextprotocol/sdk/server/index.js'
 import { CallToolRequestSchema, ListToolsRequestSchema } from '@modelcontextprotocol/sdk/types.js'
 import { net } from 'electron'
-import { z } from 'zod'
+import * as z from 'zod'
 
 const logger = loggerService.withContext('DifyKnowledgeServer')
 

src/main/mcpServers/factory.ts

@@ -3,6 +3,7 @@ import { Server } from '@modelcontextprotocol/sdk/server/index.js'
 import { BuiltinMCPServerName, BuiltinMCPServerNames } from '@types'
 
 import BraveSearchServer from './brave-search'
+import DiDiMcpServer from './didi-mcp'
 import DifyKnowledgeServer from './dify-knowledge'
 import FetchServer from './fetch'
 import FileSystemServer from './filesystem'
@@ -42,6 +43,10 @@ export function createInMemoryMCPServer(
     case BuiltinMCPServerNames.python: {
       return new PythonServer().server
     }
+    case BuiltinMCPServerNames.didiMCP: {
+      const apiKey = envs.DIDI_API_KEY
+      return new DiDiMcpServer(apiKey).server
+    }
     default:
       throw new Error(`Unknown in-memory MCP server: ${name}`)
   }

src/main/mcpServers/fetch.ts

@@ -5,7 +5,7 @@ import { CallToolRequestSchema, ListToolsRequestSchema } from '@modelcontextprot
 import { net } from 'electron'
 import { JSDOM } from 'jsdom'
 import TurndownService from 'turndown'
-import { z } from 'zod'
+import * as z from 'zod'
 
 export const RequestPayloadSchema = z.object({
   url: z.url(),

src/main/mcpServers/filesystem.ts

@@ -8,7 +8,7 @@ import fs from 'fs/promises'
 import { minimatch } from 'minimatch'
 import os from 'os'
 import path from 'path'
-import { z } from 'zod'
+import * as z from 'zod'
 
 const logger = loggerService.withContext('MCP:FileSystemServer')
 

src/main/services/ApiServerService.ts

@@ -1,5 +1,11 @@
 import { IpcChannel } from '@shared/IpcChannel'
-import { ApiServerConfig } from '@types'
+import {
+  ApiServerConfig,
+  GetApiServerStatusResult,
+  RestartApiServerStatusResult,
+  StartApiServerStatusResult,
+  StopApiServerStatusResult
+} from '@types'
 import { ipcMain } from 'electron'
 
 import { apiServer } from '../apiServer'
@@ -52,7 +58,7 @@ export class ApiServerService {
 
   registerIpcHandlers(): void {
     // API Server
-    ipcMain.handle(IpcChannel.ApiServer_Start, async () => {
+    ipcMain.handle(IpcChannel.ApiServer_Start, async (): Promise<StartApiServerStatusResult> => {
       try {
         await this.start()
         return { success: true }
@@ -61,7 +67,7 @@ export class ApiServerService {
       }
     })
 
-    ipcMain.handle(IpcChannel.ApiServer_Stop, async () => {
+    ipcMain.handle(IpcChannel.ApiServer_Stop, async (): Promise<StopApiServerStatusResult> => {
       try {
         await this.stop()
         return { success: true }
@@ -70,7 +76,7 @@ export class ApiServerService {
       }
     })
 
-    ipcMain.handle(IpcChannel.ApiServer_Restart, async () => {
+    ipcMain.handle(IpcChannel.ApiServer_Restart, async (): Promise<RestartApiServerStatusResult> => {
       try {
         await this.restart()
         return { success: true }
@@ -79,7 +85,7 @@ export class ApiServerService {
       }
     })
 
-    ipcMain.handle(IpcChannel.ApiServer_GetStatus, async () => {
+    ipcMain.handle(IpcChannel.ApiServer_GetStatus, async (): Promise<GetApiServerStatusResult> => {
       try {
         const config = await this.getCurrentConfig()
         return {

src/main/services/WebviewService.ts

@@ -1,4 +1,5 @@
-import { session, shell, webContents } from 'electron'
+import { IpcChannel } from '@shared/IpcChannel'
+import { app, session, shell, webContents } from 'electron'
 
 /**
  * init the useragent of the webview session
@@ -36,3 +37,66 @@ export function setOpenLinkExternal(webviewId: number, isExternal: boolean) {
     }
   })
 }
+
+const attachKeyboardHandler = (contents: Electron.WebContents) => {
+  if (contents.getType?.() !== 'webview') {
+    return
+  }
+
+  const handleBeforeInput = (event: Electron.Event, input: Electron.Input) => {
+    if (!input) {
+      return
+    }
+
+    const key = input.key?.toLowerCase()
+    if (!key) {
+      return
+    }
+
+    const isFindShortcut = (input.control || input.meta) && key === 'f'
+    const isEscape = key === 'escape'
+    const isEnter = key === 'enter'
+
+    if (!isFindShortcut && !isEscape && !isEnter) {
+      return
+    }
+
+    const host = contents.hostWebContents
+    if (!host || host.isDestroyed()) {
+      return
+    }
+
+    // Always prevent Cmd/Ctrl+F to override the guest page's native find dialog
+    if (isFindShortcut) {
+      event.preventDefault()
+    }
+
+    // Send the hotkey event to the renderer
+    // The renderer will decide whether to preventDefault for Escape and Enter
+    // based on whether the search bar is visible
+    host.send(IpcChannel.Webview_SearchHotkey, {
+      webviewId: contents.id,
+      key,
+      control: Boolean(input.control),
+      meta: Boolean(input.meta),
+      shift: Boolean(input.shift),
+      alt: Boolean(input.alt)
+    })
+  }
+
+  contents.on('before-input-event', handleBeforeInput)
+  contents.once('destroyed', () => {
+    contents.removeListener('before-input-event', handleBeforeInput)
+  })
+}
+
+export function initWebviewHotkeys() {
+  webContents.getAllWebContents().forEach((contents) => {
+    if (contents.isDestroyed()) return
+    attachKeyboardHandler(contents)
+  })
+
+  app.on('web-contents-created', (_, contents) => {
+    attachKeyboardHandler(contents)
+  })
+}

src/main/services/mcp/oauth/types.ts

@@ -4,7 +4,7 @@ import {
   OAuthTokens
 } from '@modelcontextprotocol/sdk/shared/auth.js'
 import EventEmitter from 'events'
-import { z } from 'zod'
+import * as z from 'zod'
 
 export interface OAuthStorageData {
   clientInfo?: OAuthClientInformation

src/main/services/ocr/OcrService.ts

@@ -2,6 +2,7 @@ import { loggerService } from '@logger'
 import { isLinux } from '@main/constant'
 import { BuiltinOcrProviderIds, OcrHandler, OcrProvider, OcrResult, SupportedOcrFile } from '@types'
 
+import { ovOcrService } from './builtin/OvOcrService'
 import { ppocrService } from './builtin/PpocrService'
 import { systemOcrService } from './builtin/SystemOcrService'
 import { tesseractService } from './builtin/TesseractService'
@@ -22,6 +23,10 @@ export class OcrService {
     this.registry.delete(providerId)
   }
 
+  public listProviderIds(): string[] {
+    return Array.from(this.registry.keys())
+  }
+
   public async ocr(file: SupportedOcrFile, provider: OcrProvider): Promise<OcrResult> {
     const handler = this.registry.get(provider.id)
     if (!handler) {
@@ -39,3 +44,5 @@ ocrService.register(BuiltinOcrProviderIds.tesseract, tesseractService.ocr.bind(t
 !isLinux && ocrService.register(BuiltinOcrProviderIds.system, systemOcrService.ocr.bind(systemOcrService))
 
 ocrService.register(BuiltinOcrProviderIds.paddleocr, ppocrService.ocr.bind(ppocrService))
+
+ovOcrService.isAvailable() && ocrService.register(BuiltinOcrProviderIds.ovocr, ovOcrService.ocr.bind(ovOcrService))

src/main/services/ocr/builtin/OvOcrService.ts

@@ -0,0 +1,128 @@
+import { loggerService } from '@logger'
+import { isWin } from '@main/constant'
+import { isImageFileMetadata, OcrOvConfig, OcrResult, SupportedOcrFile } from '@types'
+import { exec } from 'child_process'
+import * as fs from 'fs'
+import * as os from 'os'
+import * as path from 'path'
+import { promisify } from 'util'
+
+import { OcrBaseService } from './OcrBaseService'
+
+const logger = loggerService.withContext('OvOcrService')
+const execAsync = promisify(exec)
+
+const PATH_BAT_FILE = path.join(os.homedir(), '.cherrystudio', 'ovms', 'ovocr', 'run.npu.bat')
+
+export class OvOcrService extends OcrBaseService {
+  constructor() {
+    super()
+  }
+
+  public isAvailable(): boolean {
+    return (
+      isWin &&
+      os.cpus()[0].model.toLowerCase().includes('intel') &&
+      os.cpus()[0].model.toLowerCase().includes('ultra') &&
+      fs.existsSync(PATH_BAT_FILE)
+    )
+  }
+
+  private getOvOcrPath(): string {
+    return path.join(os.homedir(), '.cherrystudio', 'ovms', 'ovocr')
+  }
+
+  private getImgDir(): string {
+    return path.join(this.getOvOcrPath(), 'img')
+  }
+
+  private getOutputDir(): string {
+    return path.join(this.getOvOcrPath(), 'output')
+  }
+
+  private async clearDirectory(dirPath: string): Promise<void> {
+    if (fs.existsSync(dirPath)) {
+      const files = await fs.promises.readdir(dirPath)
+      for (const file of files) {
+        const filePath = path.join(dirPath, file)
+        const stats = await fs.promises.stat(filePath)
+        if (stats.isDirectory()) {
+          await this.clearDirectory(filePath)
+          await fs.promises.rmdir(filePath)
+        } else {
+          await fs.promises.unlink(filePath)
+        }
+      }
+    } else {
+      // If the directory does not exist, create it
+      await fs.promises.mkdir(dirPath, { recursive: true })
+    }
+  }
+
+  private async copyFileToImgDir(sourceFilePath: string, targetFileName: string): Promise<void> {
+    const imgDir = this.getImgDir()
+    const targetFilePath = path.join(imgDir, targetFileName)
+    await fs.promises.copyFile(sourceFilePath, targetFilePath)
+  }
+
+  private async runOcrBatch(): Promise<void> {
+    const ovOcrPath = this.getOvOcrPath()
+
+    try {
+      // Execute run.bat in the ov-ocr directory
+      await execAsync(`"${PATH_BAT_FILE}"`, {
+        cwd: ovOcrPath,
+        timeout: 60000 // 60 second timeout
+      })
+    } catch (error) {
+      logger.error(`Error running ovocr batch: ${error}`)
+      throw new Error(`Failed to run OCR batch: ${error}`)
+    }
+  }
+
+  private async ocrImage(filePath: string, options?: OcrOvConfig): Promise<OcrResult> {
+    logger.info(`OV OCR called on ${filePath} with options ${JSON.stringify(options)}`)
+
+    try {
+      // 1. Clear img directory and output directory
+      await this.clearDirectory(this.getImgDir())
+      await this.clearDirectory(this.getOutputDir())
+
+      // 2. Copy file to img directory
+      const fileName = path.basename(filePath)
+      await this.copyFileToImgDir(filePath, fileName)
+      logger.info(`File copied to img directory: ${fileName}`)
+
+      // 3. Run run.bat
+      logger.info('Running OV OCR batch process...')
+      await this.runOcrBatch()
+
+      // 4. Check that output/[basename].txt file exists
+      const baseNameWithoutExt = path.basename(fileName, path.extname(fileName))
+      const outputFilePath = path.join(this.getOutputDir(), `${baseNameWithoutExt}.txt`)
+      if (!fs.existsSync(outputFilePath)) {
+        throw new Error(`OV OCR output file not found at: ${outputFilePath}`)
+      }
+
+      // 5. Read output/[basename].txt file content
+      const ocrText = await fs.promises.readFile(outputFilePath, 'utf-8')
+      logger.info(`OV OCR text extracted: ${ocrText.substring(0, 100)}...`)
+
+      // 6. Return result
+      return { text: ocrText }
+    } catch (error) {
+      logger.error(`Error during OV OCR process: ${error}`)
+      throw error
+    }
+  }
+
+  public ocr = async (file: SupportedOcrFile, options?: OcrOvConfig): Promise<OcrResult> => {
+    if (isImageFileMetadata(file)) {
+      return this.ocrImage(file.path, options)
+    } else {
+      throw new Error('Unsupported file type, currently only image files are supported')
+    }
+  }
+}
+
+export const ovOcrService = new OvOcrService()

src/main/services/ocr/builtin/PpocrService.ts

@@ -1,7 +1,7 @@
 import { loadOcrImage } from '@main/utils/ocr'
 import { ImageFileMetadata, isImageFileMetadata, OcrPpocrConfig, OcrResult, SupportedOcrFile } from '@types'
 import { net } from 'electron'
-import { z } from 'zod'
+import * as z from 'zod'
 
 import { OcrBaseService } from './OcrBaseService'
 

src/main/services/remotefile/OpenAIService.ts

@@ -1,8 +1,8 @@
+import OpenAI from '@cherrystudio/openai'
 import { loggerService } from '@logger'
 import { fileStorage } from '@main/services/FileStorage'
 import { FileListResponse, FileMetadata, FileUploadResponse, Provider } from '@types'
 import * as fs from 'fs'
-import OpenAI from 'openai'
 
 import { CacheService } from '../CacheService'
 import { BaseFileService } from './BaseFileService'

src/preload/index.ts

@@ -3,7 +3,7 @@ import { SpanEntity, TokenUsage } from '@mcp-trace/trace-core'
 import { SpanContext } from '@opentelemetry/api'
 import { TerminalConfig, UpgradeChannel } from '@shared/config/constant'
 import type { LogLevel, LogSourceWithContext } from '@shared/config/logger'
-import type { FileChangeEvent } from '@shared/config/types'
+import type { FileChangeEvent, WebviewKeyEvent } from '@shared/config/types'
 import { IpcChannel } from '@shared/IpcChannel'
 import type { Notification } from '@types'
 import {
@@ -12,6 +12,7 @@ import {
   FileListResponse,
   FileMetadata,
   FileUploadResponse,
+  GetApiServerStatusResult,
   KnowledgeBaseParams,
   KnowledgeItem,
   KnowledgeSearchResult,
@@ -22,8 +23,11 @@ import {
   OcrProvider,
   OcrResult,
   Provider,
+  RestartApiServerStatusResult,
   S3Config,
   Shortcut,
+  StartApiServerStatusResult,
+  StopApiServerStatusResult,
   SupportedOcrFile,
   ThemeMode,
   WebDavConfig
@@ -390,7 +394,16 @@ const api = {
     setOpenLinkExternal: (webviewId: number, isExternal: boolean) =>
       ipcRenderer.invoke(IpcChannel.Webview_SetOpenLinkExternal, webviewId, isExternal),
     setSpellCheckEnabled: (webviewId: number, isEnable: boolean) =>
-      ipcRenderer.invoke(IpcChannel.Webview_SetSpellCheckEnabled, webviewId, isEnable)
+      ipcRenderer.invoke(IpcChannel.Webview_SetSpellCheckEnabled, webviewId, isEnable),
+    onFindShortcut: (callback: (payload: WebviewKeyEvent) => void) => {
+      const listener = (_event: Electron.IpcRendererEvent, payload: WebviewKeyEvent) => {
+        callback(payload)
+      }
+      ipcRenderer.on(IpcChannel.Webview_SearchHotkey, listener)
+      return () => {
+        ipcRenderer.off(IpcChannel.Webview_SearchHotkey, listener)
+      }
+    }
   },
   storeSync: {
     subscribe: () => ipcRenderer.invoke(IpcChannel.StoreSync_Subscribe),
@@ -467,7 +480,8 @@ const api = {
   },
   ocr: {
     ocr: (file: SupportedOcrFile, provider: OcrProvider): Promise<OcrResult> =>
-      ipcRenderer.invoke(IpcChannel.OCR_ocr, file, provider)
+      ipcRenderer.invoke(IpcChannel.OCR_ocr, file, provider),
+    listProviders: (): Promise<string[]> => ipcRenderer.invoke(IpcChannel.OCR_ListProviders)
   },
   cherryai: {
     generateSignature: (params: { method: string; path: string; query: string; body: Record<string, any> }) =>
@@ -487,6 +501,12 @@ const api = {
         ipcRenderer.removeListener(channel, listener)
       }
     }
+  },
+  apiServer: {
+    getStatus: (): Promise<GetApiServerStatusResult> => ipcRenderer.invoke(IpcChannel.ApiServer_GetStatus),
+    start: (): Promise<StartApiServerStatusResult> => ipcRenderer.invoke(IpcChannel.ApiServer_Start),
+    restart: (): Promise<RestartApiServerStatusResult> => ipcRenderer.invoke(IpcChannel.ApiServer_Restart),
+    stop: (): Promise<StopApiServerStatusResult> => ipcRenderer.invoke(IpcChannel.ApiServer_Stop)
   }
 }
 

src/renderer/src/Router.tsx

@@ -20,6 +20,7 @@ import PaintingsRoutePage from './pages/paintings/PaintingsRoutePage'
 import SettingsPage from './pages/settings/SettingsPage'
 import AssistantPresetsPage from './pages/store/assistants/presets/AssistantPresetsPage'
 import TranslatePage from './pages/translate/TranslatePage'
+import { VideoPage } from './pages/video/VideoPage'
 
 const Router: FC = () => {
   const { navbarPosition } = useNavbarPosition()
@@ -40,6 +41,7 @@ const Router: FC = () => {
           <Route path="/code" element={<CodeToolsPage />} />
           <Route path="/settings/*" element={<SettingsPage />} />
           <Route path="/launchpad" element={<LaunchpadPage />} />
+          <Route path="/video" element={<VideoPage />} />
         </Routes>
       </ErrorBoundary>
     )

src/renderer/src/aiCore/chunk/AiSdkToChunkAdapter.ts

@@ -6,6 +6,8 @@
 import { loggerService } from '@logger'
 import { AISDKWebSearchResult, MCPTool, WebSearchResults, WebSearchSource } from '@renderer/types'
 import { Chunk, ChunkType } from '@renderer/types/chunk'
+import { ProviderSpecificError } from '@renderer/types/provider-specific-error'
+import { formatErrorMessage } from '@renderer/utils/error'
 import { convertLinks, flushLinkConverterBuffer } from '@renderer/utils/linkConverter'
 import type { ClaudeCodeRawValue } from '@shared/agents/claudecode/types'
 import type { TextStreamPart, ToolSet } from 'ai'
@@ -355,7 +357,11 @@ export class AiSdkToChunkAdapter {
       case 'error':
         this.onChunk({
           type: ChunkType.ERROR,
-          error: chunk.error as Record<string, any>
+          error: new ProviderSpecificError({
+            message: formatErrorMessage(chunk.error),
+            provider: 'unknown',
+            cause: chunk.error
+          })
         })
         break
 

src/renderer/src/aiCore/index_new.ts

@@ -12,8 +12,23 @@ import { loggerService } from '@logger'
 import { getEnableDeveloperMode } from '@renderer/hooks/useSettings'
 import { addSpan, endSpan } from '@renderer/services/SpanManagerService'
 import { StartSpanParams } from '@renderer/trace/types/ModelSpanEntity'
-import type { Assistant, GenerateImageParams, Model, Provider } from '@renderer/types'
+import type {
+  Assistant,
+  DeleteVideoParams,
+  DeleteVideoResult,
+  GenerateImageParams,
+  Model,
+  Provider,
+  RetrieveVideoContentParams
+} from '@renderer/types'
 import type { AiSdkModel, StreamTextParams } from '@renderer/types/aiCoreTypes'
+import {
+  CreateVideoParams,
+  CreateVideoResult,
+  RetrieveVideoContentResult,
+  RetrieveVideoParams,
+  RetrieveVideoResult
+} from '@renderer/types/video'
 import { buildClaudeCodeSystemModelMessage } from '@shared/anthropic'
 import { type ImageModel, type LanguageModel, type Provider as AiSdkProvider, wrapLanguageModel } from 'ai'
 
@@ -83,10 +98,8 @@ export default class ModernAiProvider {
       throw new Error('Model is required for completions. Please use constructor with model parameter.')
     }
 
-    // 确保配置存在
-    if (!this.config) {
-      this.config = providerToAiSdkConfig(this.actualProvider, this.model)
-    }
+    // 每次请求时重新生成配置以确保API key轮换生效
+    this.config = providerToAiSdkConfig(this.actualProvider, this.model)
 
     // 准备特殊配置
     await prepareSpecialProviderConfig(this.actualProvider, this.config)
@@ -500,6 +513,34 @@ export default class ModernAiProvider {
     return images
   }
 
+  /**
+   * We manually implement this method before aisdk supports it well
+   */
+  public async createVideo(params: CreateVideoParams): Promise<CreateVideoResult> {
+    return this.legacyProvider.createVideo(params)
+  }
+
+  /**
+   * We manually implement this method before aisdk supports it well
+   */
+  public async retrieveVideo(params: RetrieveVideoParams): Promise<RetrieveVideoResult> {
+    return this.legacyProvider.retrieveVideo(params)
+  }
+
+  /**
+   * We manually implement this method before aisdk supports it well
+   */
+  public async retrieveVideoContent(params: RetrieveVideoContentParams): Promise<RetrieveVideoContentResult> {
+    return this.legacyProvider.retrieveVideoContent(params)
+  }
+
+  /**
+   * We manually implement this method before aisdk supports it well
+   */
+  public async deleteVideo(params: DeleteVideoParams): Promise<DeleteVideoResult> {
+    return this.legacyProvider.deleteVideo(params)
+  }
+
   public getBaseURL(): string {
     return this.legacyProvider.getBaseURL()
   }

src/renderer/src/aiCore/legacy/clients/BaseApiClient.ts

@@ -70,13 +70,19 @@ export abstract class BaseApiClient<
 {
   public provider: Provider
   protected host: string
-  protected apiKey: string
   protected sdkInstance?: TSdkInstance
 
   constructor(provider: Provider) {
     this.provider = provider
     this.host = this.getBaseURL()
-    this.apiKey = this.getApiKey()
+  }
+
+  /**
+   * Get the current API key with rotation support
+   * This getter ensures API keys rotate on each access when multiple keys are configured
+   */
+  protected get apiKey(): string {
+    return this.getApiKey()
   }
 
   /**

src/renderer/src/aiCore/legacy/clients/cherryai/CherryAiAPIClient.ts

@@ -1,6 +1,6 @@
+import OpenAI from '@cherrystudio/openai'
 import { Provider } from '@renderer/types'
 import { OpenAISdkParams, OpenAISdkRawOutput } from '@renderer/types/sdk'
-import OpenAI from 'openai'
 
 import { OpenAIAPIClient } from '../openai/OpenAIApiClient'
 

src/renderer/src/aiCore/legacy/clients/openai/OpenAIApiClient.ts

@@ -1,3 +1,9 @@
+import OpenAI, { AzureOpenAI } from '@cherrystudio/openai'
+import {
+  ChatCompletionContentPart,
+  ChatCompletionContentPartRefusal,
+  ChatCompletionTool
+} from '@cherrystudio/openai/resources'
 import { loggerService } from '@logger'
 import { DEFAULT_MAX_TOKENS } from '@renderer/config/constant'
 import {
@@ -78,8 +84,6 @@ import {
 } from '@renderer/utils/mcp-tools'
 import { findFileBlocks, findImageBlocks } from '@renderer/utils/messageUtils/find'
 import { t } from 'i18next'
-import OpenAI, { AzureOpenAI } from 'openai'
-import { ChatCompletionContentPart, ChatCompletionContentPartRefusal, ChatCompletionTool } from 'openai/resources'
 
 import { GenericChunk } from '../../middleware/schemas'
 import { RequestTransformer, ResponseChunkTransformer, ResponseChunkTransformerContext } from '../types'

src/renderer/src/aiCore/legacy/clients/openai/OpenAIBaseClient.ts

@@ -1,4 +1,6 @@
+import OpenAI, { AzureOpenAI } from '@cherrystudio/openai'
 import { loggerService } from '@logger'
+import { COPILOT_DEFAULT_HEADERS } from '@renderer/aiCore/provider/constants'
 import {
   isClaudeReasoningModel,
   isOpenAIReasoningModel,
@@ -24,7 +26,6 @@ import {
   ReasoningEffortOptionalParams
 } from '@renderer/types/sdk'
 import { formatApiHost } from '@renderer/utils/api'
-import OpenAI, { AzureOpenAI } from 'openai'
 
 import { BaseApiClient } from '../BaseApiClient'
 
@@ -166,7 +167,8 @@ export abstract class OpenAIBaseClient<
         baseURL: this.getBaseURL(),
         defaultHeaders: {
           ...this.defaultHeaders(),
-          ...this.provider.extra_headers
+          ...this.provider.extra_headers,
+          ...(this.provider.id === 'copilot' ? COPILOT_DEFAULT_HEADERS : {})
         }
       }) as TSdkInstance
     }

src/renderer/src/aiCore/legacy/clients/openai/OpenAIResponseAPIClient.ts

@@ -1,3 +1,5 @@
+import OpenAI, { AzureOpenAI } from '@cherrystudio/openai'
+import { ResponseInput } from '@cherrystudio/openai/resources/responses/responses'
 import { loggerService } from '@logger'
 import { GenericChunk } from '@renderer/aiCore/legacy/middleware/schemas'
 import { CompletionsContext } from '@renderer/aiCore/legacy/middleware/types'
@@ -34,6 +36,12 @@ import {
   OpenAIResponseSdkTool,
   OpenAIResponseSdkToolCall
 } from '@renderer/types/sdk'
+import {
+  CreateVideoParams,
+  DeleteVideoParams,
+  RetrieveVideoContentParams,
+  RetrieveVideoParams
+} from '@renderer/types/video'
 import { addImageFileToContents } from '@renderer/utils/formats'
 import {
   isSupportedToolUse,
@@ -45,8 +53,6 @@ import { findFileBlocks, findImageBlocks } from '@renderer/utils/messageUtils/fi
 import { MB } from '@shared/config/constant'
 import { t } from 'i18next'
 import { isEmpty } from 'lodash'
-import OpenAI, { AzureOpenAI } from 'openai'
-import { ResponseInput } from 'openai/resources/responses/responses'
 
 import { RequestTransformer, ResponseChunkTransformer } from '../types'
 import { OpenAIAPIClient } from './OpenAIApiClient'
@@ -152,6 +158,26 @@ export class OpenAIResponseAPIClient extends OpenAIBaseClient<
     return await sdk.responses.create(payload, options)
   }
 
+  public async createVideo(params: CreateVideoParams): Promise<OpenAI.Videos.Video> {
+    const sdk = await this.getSdkInstance()
+    return sdk.videos.create(params.params, params.options)
+  }
+
+  public async retrieveVideo(params: RetrieveVideoParams): Promise<OpenAI.Videos.Video> {
+    const sdk = await this.getSdkInstance()
+    return sdk.videos.retrieve(params.videoId, params.options)
+  }
+
+  public async retrieveVideoContent(params: RetrieveVideoContentParams): Promise<Response> {
+    const sdk = await this.getSdkInstance()
+    return sdk.videos.downloadContent(params.videoId, params.query, params.options)
+  }
+
+  public async deleteVideo(params: DeleteVideoParams): Promise<OpenAI.Videos.VideoDeleteResponse> {
+    const sdk = await this.getSdkInstance()
+    return sdk.videos.delete(params.videoId, params.options)
+  }
+
   private async handlePdfFile(file: FileMetadata): Promise<OpenAI.Responses.ResponseInputFile | undefined> {
     if (file.size > 32 * MB) return undefined
     try {
@@ -343,7 +369,14 @@ export class OpenAIResponseAPIClient extends OpenAIBaseClient<
     }
     switch (message.type) {
       case 'function_call_output':
-        sum += estimateTextTokens(message.output)
+        if (typeof message.output === 'string') {
+          sum += estimateTextTokens(message.output)
+        } else {
+          sum += message.output
+            .filter((item) => item.type === 'input_text')
+            .map((item) => estimateTextTokens(item.text))
+            .reduce((prev, cur) => prev + cur, 0)
+        }
         break
       case 'function_call':
         sum += estimateTextTokens(message.arguments)

src/renderer/src/aiCore/legacy/clients/ovms/OVMSClient.ts

@@ -1,7 +1,7 @@
+import OpenAI from '@cherrystudio/openai'
 import { loggerService } from '@logger'
 import { isSupportedModel } from '@renderer/config/models'
 import { objectKeys, Provider } from '@renderer/types'
-import OpenAI from 'openai'
 
 import { OpenAIAPIClient } from '../openai/OpenAIApiClient'
 

src/renderer/src/aiCore/legacy/clients/ppio/PPIOAPIClient.ts

@@ -1,7 +1,7 @@
+import OpenAI from '@cherrystudio/openai'
 import { loggerService } from '@logger'
 import { isSupportedModel } from '@renderer/config/models'
 import { Model, Provider } from '@renderer/types'
-import OpenAI from 'openai'
 
 import { OpenAIAPIClient } from '../openai/OpenAIApiClient'
 

src/renderer/src/aiCore/legacy/clients/types.ts

@@ -1,4 +1,5 @@
 import Anthropic from '@anthropic-ai/sdk'
+import OpenAI from '@cherrystudio/openai'
 import { Assistant, MCPTool, MCPToolResponse, Model, ToolCallResponse } from '@renderer/types'
 import { Provider } from '@renderer/types'
 import {
@@ -13,7 +14,6 @@ import {
   SdkTool,
   SdkToolCall
 } from '@renderer/types/sdk'
-import OpenAI from 'openai'
 
 import { CompletionsParams, GenericChunk } from '../middleware/schemas'
 import { CompletionsContext } from '../middleware/types'

src/renderer/src/aiCore/legacy/clients/zhipu/ZhipuAPIClient.ts

@@ -1,7 +1,7 @@
+import OpenAI from '@cherrystudio/openai'
 import { loggerService } from '@logger'
 import { Provider } from '@renderer/types'
 import { GenerateImageParams } from '@renderer/types'
-import OpenAI from 'openai'
 
 import { OpenAIAPIClient } from '../openai/OpenAIApiClient'
 

src/renderer/src/aiCore/legacy/index.ts

@@ -5,8 +5,22 @@ import { isDedicatedImageGenerationModel, isFunctionCallingModel } from '@render
 import { getProviderByModel } from '@renderer/services/AssistantService'
 import { withSpanResult } from '@renderer/services/SpanManagerService'
 import { StartSpanParams } from '@renderer/trace/types/ModelSpanEntity'
-import type { GenerateImageParams, Model, Provider } from '@renderer/types'
+import type {
+  DeleteVideoParams,
+  DeleteVideoResult,
+  GenerateImageParams,
+  Model,
+  Provider,
+  RetrieveVideoContentParams
+} from '@renderer/types'
 import type { RequestOptions, SdkModel } from '@renderer/types/sdk'
+import {
+  CreateVideoParams,
+  CreateVideoResult,
+  RetrieveVideoContentResult,
+  RetrieveVideoParams,
+  RetrieveVideoResult
+} from '@renderer/types/video'
 import { isSupportedToolUse } from '@renderer/utils/mcp-tools'
 
 import { AihubmixAPIClient } from './clients/aihubmix/AihubmixAPIClient'
@@ -179,6 +193,54 @@ export default class AiProvider {
     return this.apiClient.generateImage(params)
   }
 
+  public async createVideo(params: CreateVideoParams): Promise<CreateVideoResult> {
+    if (this.apiClient instanceof OpenAIResponseAPIClient && params.type === 'openai') {
+      const video = await this.apiClient.createVideo(params)
+      return {
+        type: 'openai',
+        video
+      }
+    } else {
+      throw new Error('Video generation is not supported by this provider')
+    }
+  }
+
+  public async retrieveVideo(params: RetrieveVideoParams): Promise<RetrieveVideoResult> {
+    if (this.apiClient instanceof OpenAIResponseAPIClient && params.type === 'openai') {
+      const video = await this.apiClient.retrieveVideo(params)
+      return {
+        type: 'openai',
+        video
+      }
+    } else {
+      throw new Error('Video generation is not supported by this provider')
+    }
+  }
+
+  public async retrieveVideoContent(params: RetrieveVideoContentParams): Promise<RetrieveVideoContentResult> {
+    if (this.apiClient instanceof OpenAIResponseAPIClient && params.type === 'openai') {
+      const response = await this.apiClient.retrieveVideoContent(params)
+      return {
+        type: 'openai',
+        response
+      }
+    } else {
+      throw new Error('Video generation is not supported by this provider')
+    }
+  }
+
+  public async deleteVideo(params: DeleteVideoParams): Promise<DeleteVideoResult> {
+    if (this.apiClient instanceof OpenAIResponseAPIClient && params.type === 'openai') {
+      const result = await this.apiClient.deleteVideo(params)
+      return {
+        type: 'openai',
+        result
+      }
+    } else {
+      throw new Error('Video deletion is not supported by this provider')
+    }
+  }
+
   public getBaseURL(): string {
     return this.apiClient.getBaseURL()
   }

src/renderer/src/aiCore/legacy/middleware/feat/ImageGenerationMiddleware.ts

@@ -1,10 +1,10 @@
+import OpenAI from '@cherrystudio/openai'
+import { toFile } from '@cherrystudio/openai/uploads'
 import { isDedicatedImageGenerationModel } from '@renderer/config/models'
 import FileManager from '@renderer/services/FileManager'
 import { ChunkType } from '@renderer/types/chunk'
 import { findImageBlocks, getMainTextContent } from '@renderer/utils/messageUtils/find'
 import { defaultTimeout } from '@shared/config/constant'
-import OpenAI from 'openai'
-import { toFile } from 'openai/uploads'
 
 import { BaseApiClient } from '../../clients/BaseApiClient'
 import { CompletionsParams, CompletionsResult, GenericChunk } from '../schemas'

src/renderer/src/aiCore/middleware/AiSdkMiddlewareBuilder.ts

@@ -4,6 +4,8 @@ import type { MCPTool, Message, Model, Provider } from '@renderer/types'
 import type { Chunk } from '@renderer/types/chunk'
 import { extractReasoningMiddleware, LanguageModelMiddleware, simulateStreamingMiddleware } from 'ai'
 
+import { noThinkMiddleware } from './noThinkMiddleware'
+
 const logger = loggerService.withContext('AiSdkMiddlewareBuilder')
 
 /**
@@ -186,6 +188,14 @@ function addProviderSpecificMiddlewares(builder: AiSdkMiddlewareBuilder, config:
       // 其他provider的通用处理
       break
   }
+
+  // OVMS+MCP's specific middleware
+  if (config.provider.id === 'ovms' && config.mcpTools && config.mcpTools.length > 0) {
+    builder.add({
+      name: 'no-think',
+      middleware: noThinkMiddleware()
+    })
+  }
 }
 
 /**

src/renderer/src/aiCore/middleware/noThinkMiddleware.ts

@@ -0,0 +1,52 @@
+import { loggerService } from '@logger'
+import { LanguageModelMiddleware } from 'ai'
+
+const logger = loggerService.withContext('noThinkMiddleware')
+
+/**
+ * No Think Middleware
+ * Automatically appends ' /no_think' string to the end of user messages for the provider
+ * This prevents the model from generating unnecessary thinking process and returns results directly
+ * @returns LanguageModelMiddleware
+ */
+export function noThinkMiddleware(): LanguageModelMiddleware {
+  return {
+    middlewareVersion: 'v2',
+
+    transformParams: async ({ params }) => {
+      const transformedParams = { ...params }
+      // Process messages in prompt
+      if (transformedParams.prompt && Array.isArray(transformedParams.prompt)) {
+        transformedParams.prompt = transformedParams.prompt.map((message) => {
+          // Only process user messages
+          if (message.role === 'user') {
+            // Process content array
+            if (Array.isArray(message.content)) {
+              const lastContent = message.content[message.content.length - 1]
+              // If the last content is text type, append ' /no_think'
+              if (lastContent && lastContent.type === 'text' && typeof lastContent.text === 'string') {
+                // Avoid duplicate additions
+                if (!lastContent.text.endsWith('/no_think')) {
+                  logger.debug('Adding /no_think to user message')
+                  return {
+                    ...message,
+                    content: [
+                      ...message.content.slice(0, -1),
+                      {
+                        ...lastContent,
+                        text: lastContent.text + ' /no_think'
+                      }
+                    ]
+                  }
+                }
+              }
+            }
+          }
+          return message
+        })
+      }
+
+      return transformedParams
+    }
+  }
+}

src/renderer/src/aiCore/prepareParams/fileProcessor.ts

@@ -3,6 +3,7 @@
  * 处理文件内容提取、文件格式转换、文件上传等逻辑
  */
 
+import type OpenAI from '@cherrystudio/openai'
 import { loggerService } from '@logger'
 import { getProviderByModel } from '@renderer/services/AssistantService'
 import type { FileMetadata, Message, Model } from '@renderer/types'
@@ -10,7 +11,6 @@ import { FileTypes } from '@renderer/types'
 import { FileMessageBlock } from '@renderer/types/newMessage'
 import { findFileBlocks } from '@renderer/utils/messageUtils/find'
 import type { FilePart, TextPart } from 'ai'
-import type OpenAI from 'openai'
 
 import { getAiSdkProviderId } from '../provider/factory'
 import { getFileSizeLimit, supportsImageInput, supportsLargeFileUpload, supportsPdfInput } from './modelCapabilities'

src/renderer/src/aiCore/provider/providerConfig.ts

@@ -63,13 +63,14 @@ function handleSpecialProviders(model: Model, provider: Provider): Provider {
   //   return createVertexProvider(provider)
   // }
 
+  if (isNewApiProvider(provider)) {
+    return newApiResolverCreator(model, provider)
+  }
+
   if (isSystemProvider(provider)) {
     if (provider.id === 'aihubmix') {
       return aihubmixProviderCreator(model, provider)
     }
-    if (isNewApiProvider(provider)) {
-      return newApiResolverCreator(model, provider)
-    }
     if (provider.id === 'vertexai') {
       return vertexAnthropicProviderCreator(model, provider)
     }

src/renderer/src/aiCore/provider/providerInitialization.ts

@@ -55,6 +55,14 @@ export const NEW_PROVIDER_CONFIGS: ProviderConfig[] = [
     creatorFunctionName: 'createPerplexity',
     supportsImageGeneration: false,
     aliases: ['perplexity']
+  },
+  {
+    id: 'mistral',
+    name: 'Mistral',
+    import: () => import('@ai-sdk/mistral'),
+    creatorFunctionName: 'createMistral',
+    supportsImageGeneration: false,
+    aliases: ['mistral']
   }
 ] as const
 

src/renderer/src/aiCore/tools/KnowledgeSearchTool.ts

@@ -4,7 +4,7 @@ import type { Assistant, KnowledgeReference } from '@renderer/types'
 import { ExtractResults, KnowledgeExtractResults } from '@renderer/utils/extract'
 import { type InferToolInput, type InferToolOutput, tool } from 'ai'
 import { isEmpty } from 'lodash'
-import { z } from 'zod'
+import * as z from 'zod'
 
 /**
  * 知识库搜索工具

src/renderer/src/aiCore/tools/MemorySearchTool.ts

@@ -1,7 +1,7 @@
 import store from '@renderer/store'
 import { selectCurrentUserId, selectGlobalMemoryEnabled, selectMemoryConfig } from '@renderer/store/memory'
 import { type InferToolInput, type InferToolOutput, tool } from 'ai'
-import { z } from 'zod'
+import * as z from 'zod'
 
 import { MemoryProcessor } from '../../services/MemoryProcessor'
 

src/renderer/src/aiCore/tools/WebSearchTool.ts

@@ -3,7 +3,7 @@ import WebSearchService from '@renderer/services/WebSearchService'
 import { WebSearchProvider, WebSearchProviderResponse } from '@renderer/types'
 import { ExtractResults } from '@renderer/utils/extract'
 import { type InferToolInput, type InferToolOutput, tool } from 'ai'
-import { z } from 'zod'
+import * as z from 'zod'
 
 /**
  * 使用预提取关键词的网络搜索工具

src/renderer/src/aiCore/utils/reasoning.ts

@@ -6,6 +6,7 @@ import {
   getThinkModelType,
   isDeepSeekHybridInferenceModel,
   isDoubaoThinkingAutoModel,
+  isGrok4FastReasoningModel,
   isGrokReasoningModel,
   isOpenAIReasoningModel,
   isQwenAlwaysThinkModel,
@@ -52,7 +53,12 @@ export function getReasoningEffort(assistant: Assistant, model: Model): Reasonin
         return {}
       }
       // Don't disable reasoning for models that require it
-      if (isGrokReasoningModel(model) || isOpenAIReasoningModel(model) || model.id.includes('seed-oss')) {
+      if (
+        isGrokReasoningModel(model) ||
+        isOpenAIReasoningModel(model) ||
+        isQwenAlwaysThinkModel(model) ||
+        model.id.includes('seed-oss')
+      ) {
         return {}
       }
       return { reasoning: { enabled: false, exclude: true } }
@@ -100,6 +106,7 @@ export function getReasoningEffort(assistant: Assistant, model: Model): Reasonin
   // reasoningEffort有效的情况
   // DeepSeek hybrid inference models, v3.1 and maybe more in the future
   // 不同的 provider 有不同的思考控制方式，在这里统一解决
+
   if (isDeepSeekHybridInferenceModel(model)) {
     if (isSystemProvider(provider)) {
       switch (provider.id) {
@@ -142,6 +149,16 @@ export function getReasoningEffort(assistant: Assistant, model: Model): Reasonin
 
   // OpenRouter models
   if (model.provider === SystemProviderIds.openrouter) {
+    // Grok 4 Fast doesn't support effort levels, always use enabled: true
+    if (isGrok4FastReasoningModel(model)) {
+      return {
+        reasoning: {
+          enabled: true // Ignore effort level, just enable reasoning
+        }
+      }
+    }
+
+    // Other OpenRouter models that support effort levels
     if (isSupportedReasoningEffortModel(model) || isSupportedThinkingTokenModel(model)) {
       return {
         reasoning: {
@@ -412,6 +429,13 @@ export function getGeminiReasoningParams(assistant: Assistant, model: Model): Re
   return {}
 }
 
+/**
+ * Get XAI-specific reasoning parameters
+ * This function should only be called for XAI provider models
+ * @param assistant - The assistant configuration
+ * @param model - The model being used
+ * @returns XAI-specific reasoning parameters
+ */
 export function getXAIReasoningParams(assistant: Assistant, model: Model): Record<string, any> {
   if (!isSupportedReasoningEffortGrokModel(model)) {
     return {}
@@ -419,6 +443,11 @@ export function getXAIReasoningParams(assistant: Assistant, model: Model): Recor
 
   const { reasoning_effort: reasoningEffort } = getAssistantSettings(assistant)
 
+  if (!reasoningEffort) {
+    return {}
+  }
+
+  // For XAI provider Grok models, use reasoningEffort parameter directly
   return {
     reasoningEffort
   }

src/renderer/src/api/agent.ts

@@ -9,6 +9,8 @@ import {
   CreateAgentRequest,
   CreateAgentResponse,
   CreateAgentResponseSchema,
+  CreateAgentSessionResponse,
+  CreateAgentSessionResponseSchema,
   CreateSessionForm,
   CreateSessionRequest,
   GetAgentResponse,
@@ -171,12 +173,12 @@ export class AgentApiClient {
     }
   }
 
-  public async createSession(agentId: string, session: CreateSessionForm): Promise<GetAgentSessionResponse> {
+  public async createSession(agentId: string, session: CreateSessionForm): Promise<CreateAgentSessionResponse> {
     const url = this.getSessionPaths(agentId).base
     try {
       const payload = session satisfies CreateSessionRequest
       const response = await this.axios.post(url, payload)
-      const data = GetAgentSessionResponseSchema.parse(response.data)
+      const data = CreateAgentSessionResponseSchema.parse(response.data)
       return data
     } catch (error) {
       throw processError(error, 'Failed to add session.')

src/renderer/src/components/CodeBlockView/view.tsx

@@ -326,7 +326,7 @@ const CodeBlockWrapper = styled.div<{ $isInSpecialView: boolean }>`
    * 一是 CodeViewer 在气泡样式下的用户消息中无法撑开气泡，
    * 二是 代码块内容过少时 toolbar 会和 title 重叠。
    */
-  min-width: 45ch;
+  min-width: 35ch;
 
   .code-toolbar {
     background-color: ${(props) => (props.$isInSpecialView ? 'transparent' : 'var(--color-background-mute)')};

src/renderer/src/components/CodeEditor/hooks.ts

@@ -2,7 +2,7 @@ import { linter } from '@codemirror/lint' // statically imported by @uiw/codemir
 import { EditorView } from '@codemirror/view'
 import { loggerService } from '@logger'
 import { Extension, keymap } from '@uiw/react-codemirror'
-import { useEffect, useMemo, useState } from 'react'
+import { useCallback, useEffect, useMemo, useState } from 'react'
 
 import { getNormalizedExtension } from './utils'
 
@@ -203,3 +203,80 @@ export function useHeightListener({ onHeightChange }: UseHeightListenerProps) {
     })
   }, [onHeightChange])
 }
+
+interface UseScrollToLineOptions {
+  highlight?: boolean
+}
+
+export function useScrollToLine(editorViewRef: React.MutableRefObject<EditorView | null>) {
+  const findLineElement = useCallback((view: EditorView, position: number): HTMLElement | null => {
+    const domAtPos = view.domAtPos(position)
+    let node: Node | null = domAtPos.node
+
+    if (node.nodeType === Node.TEXT_NODE) {
+      node = node.parentElement
+    }
+
+    while (node) {
+      if (node instanceof HTMLElement && node.classList.contains('cm-line')) {
+        return node
+      }
+      node = node.parentElement
+    }
+
+    return null
+  }, [])
+
+  const highlightLine = useCallback((view: EditorView, element: HTMLElement) => {
+    const previousHighlight = view.dom.querySelector('.animation-locate-highlight') as HTMLElement | null
+    if (previousHighlight) {
+      previousHighlight.classList.remove('animation-locate-highlight')
+    }
+
+    element.classList.add('animation-locate-highlight')
+
+    const handleAnimationEnd = () => {
+      element.classList.remove('animation-locate-highlight')
+      element.removeEventListener('animationend', handleAnimationEnd)
+    }
+
+    element.addEventListener('animationend', handleAnimationEnd)
+  }, [])
+
+  return useCallback(
+    (lineNumber: number, options?: UseScrollToLineOptions) => {
+      const view = editorViewRef.current
+      if (!view) return
+
+      const targetLine = view.state.doc.line(Math.min(lineNumber, view.state.doc.lines))
+
+      const lineElement = findLineElement(view, targetLine.from)
+      if (lineElement) {
+        lineElement.scrollIntoView({ behavior: 'smooth', block: 'center', inline: 'nearest' })
+
+        if (options?.highlight) {
+          requestAnimationFrame(() => highlightLine(view, lineElement))
+        }
+        return
+      }
+
+      view.dispatch({
+        effects: EditorView.scrollIntoView(targetLine.from, {
+          y: 'start'
+        })
+      })
+
+      if (!options?.highlight) {
+        return
+      }
+
+      setTimeout(() => {
+        const fallbackElement = findLineElement(view, targetLine.from)
+        if (fallbackElement) {
+          highlightLine(view, fallbackElement)
+        }
+      }, 200)
+    },
+    [editorViewRef, findLineElement, highlightLine]
+  )
+}

src/renderer/src/components/CodeEditor/index.tsx

@@ -5,13 +5,14 @@ import diff from 'fast-diff'
 import { useCallback, useEffect, useImperativeHandle, useMemo, useRef } from 'react'
 import { memo } from 'react'
 
-import { useBlurHandler, useHeightListener, useLanguageExtensions, useSaveKeymap } from './hooks'
+import { useBlurHandler, useHeightListener, useLanguageExtensions, useSaveKeymap, useScrollToLine } from './hooks'
 
 // 标记非用户编辑的变更
 const External = Annotation.define<boolean>()
 
 export interface CodeEditorHandles {
   save?: () => void
+  scrollToLine?: (lineNumber: number, options?: { highlight?: boolean }) => void
 }
 
 export interface CodeEditorProps {
@@ -181,8 +182,11 @@ const CodeEditor = ({
     ].flat()
   }, [extensions, langExtensions, wrapped, saveKeymapExtension, blurExtension, heightListenerExtension])
 
+  const scrollToLine = useScrollToLine(editorViewRef)
+
   useImperativeHandle(ref, () => ({
-    save: handleSave
+    save: handleSave,
+    scrollToLine
   }))
 
   return (

src/renderer/src/components/HighlightText.tsx

@@ -0,0 +1,48 @@
+import { FC, memo, useMemo } from 'react'
+
+interface HighlightTextProps {
+  text: string
+  keyword: string
+  caseSensitive?: boolean
+  className?: string
+}
+
+/**
+ * Text highlighting component that marks keyword matches
+ */
+const HighlightText: FC<HighlightTextProps> = ({ text, keyword, caseSensitive = false, className }) => {
+  const highlightedText = useMemo(() => {
+    if (!keyword || !text) {
+      return <span>{text}</span>
+    }
+
+    // Escape regex special characters
+    const escapedKeyword = keyword.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')
+    const flags = caseSensitive ? 'g' : 'gi'
+    const regex = new RegExp(`(${escapedKeyword})`, flags)
+
+    // Split text by keyword matches
+    const parts = text.split(regex)
+
+    return (
+      <>
+        {parts.map((part, index) => {
+          // Check if part matches keyword
+          const isMatch = regex.test(part)
+          regex.lastIndex = 0 // Reset regex state
+
+          if (isMatch) {
+            return <mark key={index}>{part}</mark>
+          }
+          return <span key={index}>{part}</span>
+        })}
+      </>
+    )
+  }, [text, keyword, caseSensitive])
+
+  const combinedClassName = className ? `ant-typography ${className}` : 'ant-typography'
+
+  return <span className={combinedClassName}>{highlightedText}</span>
+}
+
+export default memo(HighlightText)

src/renderer/src/components/MinApp/WebviewContainer.tsx

@@ -25,7 +25,7 @@ const WebviewContainer = memo(
     onNavigateCallback: (appid: string, url: string) => void
   }) => {
     const webviewRef = useRef<WebviewTag | null>(null)
-    const { enableSpellCheck } = useSettings()
+    const { enableSpellCheck, minappsOpenLinkExternal } = useSettings()
 
     const setRef = (appid: string) => {
       onSetRefCallback(appid, null)
@@ -76,6 +76,8 @@ const WebviewContainer = memo(
         const webviewId = webviewRef.current?.getWebContentsId()
         if (webviewId) {
           window.api?.webview?.setSpellCheckEnabled?.(webviewId, enableSpellCheck)
+          // Set link opening behavior for this webview
+          window.api?.webview?.setOpenLinkExternal?.(webviewId, minappsOpenLinkExternal)
         }
       }
 
@@ -104,6 +106,22 @@ const WebviewContainer = memo(
       // eslint-disable-next-line react-hooks/exhaustive-deps
     }, [appid, url])
 
+    // Update webview settings when they change
+    useEffect(() => {
+      if (!webviewRef.current) return
+
+      try {
+        const webviewId = webviewRef.current.getWebContentsId()
+        if (webviewId) {
+          window.api?.webview?.setSpellCheckEnabled?.(webviewId, enableSpellCheck)
+          window.api?.webview?.setOpenLinkExternal?.(webviewId, minappsOpenLinkExternal)
+        }
+      } catch (error) {
+        // WebView may not be ready yet, settings will be applied in dom-ready event
+        logger.debug(`WebView ${appid} not ready for settings update`)
+      }
+    }, [appid, minappsOpenLinkExternal, enableSpellCheck])
+
     const WebviewStyle: React.CSSProperties = {
       width: '100%',
       height: '100%',

src/renderer/src/components/Popups/agent/AgentModal.tsx

@@ -17,7 +17,7 @@ import {
 import { loggerService } from '@logger'
 import type { Selection } from '@react-types/shared'
 import ClaudeIcon from '@renderer/assets/images/models/claude.png'
-import { getModelLogo } from '@renderer/config/models'
+import { agentModelFilter, getModelLogo } from '@renderer/config/models'
 import { permissionModeCards } from '@renderer/constants/permissionModes'
 import { useAgents } from '@renderer/hooks/agents/useAgents'
 import { useApiModels } from '@renderer/hooks/agents/useModels'
@@ -245,14 +245,23 @@ export const AgentModal: React.FC<Props> = ({ agent, trigger, isOpen: _isOpen, o
 
   const modelOptions = useMemo(() => {
     // mocked data. not final version
-    return (models ?? []).map((model) => ({
-      type: 'model',
-      key: model.id,
-      label: model.name,
-      avatar: getModelLogo(model.id),
-      providerId: model.provider,
-      providerName: model.provider_name
-    })) satisfies ModelOption[]
+    return (models ?? [])
+      .filter((m) =>
+        agentModelFilter({
+          id: m.id,
+          provider: m.provider || '',
+          name: m.name,
+          group: ''
+        })
+      )
+      .map((model) => ({
+        type: 'model',
+        key: model.id,
+        label: model.name,
+        avatar: getModelLogo(model.id),
+        providerId: model.provider,
+        providerName: model.provider_name
+      })) satisfies ModelOption[]
   }, [models])
 
   const onModelChange = useCallback((e: ChangeEvent<HTMLSelectElement>) => {

src/renderer/src/components/Popups/agent/SessionModal.tsx

@@ -98,7 +98,7 @@ export const SessionModal: React.FC<Props> = ({
   const loadingRef = useRef(false)
   // const { setTimeoutTimer } = useTimer()
   const { createSession } = useSessions(agentId)
-  const updateSession = useUpdateSession(agentId)
+  const { updateSession } = useUpdateSession(agentId)
   const { agent } = useAgent(agentId)
   const isEditing = (session?: AgentSessionEntity) => session !== undefined
 

src/renderer/src/components/RichEditor/constants.ts

@@ -0,0 +1,2 @@
+// Attribute used to store the original source line number in markdown editors
+export const MARKDOWN_SOURCE_LINE_ATTR = 'data-source-line'

src/renderer/src/components/RichEditor/index.tsx

@@ -1,5 +1,6 @@
 import { loggerService } from '@logger'
 import { ContentSearch, type ContentSearchRef } from '@renderer/components/ContentSearch'
+import { MARKDOWN_SOURCE_LINE_ATTR } from '@renderer/components/RichEditor/constants'
 import DragHandle from '@tiptap/extension-drag-handle-react'
 import { EditorContent } from '@tiptap/react'
 import { Tooltip } from 'antd'
@@ -29,6 +30,156 @@ import type { FormattingCommand, RichEditorProps, RichEditorRef } from './types'
 import { useRichEditor } from './useRichEditor'
 const logger = loggerService.withContext('RichEditor')
 
+/**
+ * Find element by line number with fallback strategies:
+ * 1. Exact line + content match
+ * 2. Exact line match
+ * 3. Closest line <= target
+ */
+function findElementByLine(editorDom: HTMLElement, lineNumber: number, lineContent?: string): HTMLElement | null {
+  const allElements = Array.from(editorDom.querySelectorAll(`[${MARKDOWN_SOURCE_LINE_ATTR}]`)) as HTMLElement[]
+  if (allElements.length === 0) {
+    logger.warn('No elements with data-source-line attribute found')
+    return null
+  }
+  const exactMatches = editorDom.querySelectorAll(
+    `[${MARKDOWN_SOURCE_LINE_ATTR}="${lineNumber}"]`
+  ) as NodeListOf<HTMLElement>
+
+  // Strategy 1: Exact line + content match
+  if (exactMatches.length > 1 && lineContent) {
+    for (const match of Array.from(exactMatches)) {
+      if (match.textContent?.includes(lineContent)) {
+        return match
+      }
+    }
+  }
+
+  // Strategy 2: Exact line match
+  if (exactMatches.length > 0) {
+    return exactMatches[0]
+  }
+
+  // Strategy 3: Closest line <= target
+  let closestElement: HTMLElement | null = null
+  let closestLine = 0
+
+  for (const el of allElements) {
+    const sourceLine = parseInt(el.getAttribute(MARKDOWN_SOURCE_LINE_ATTR) || '0', 10)
+    if (sourceLine <= lineNumber && sourceLine > closestLine) {
+      closestLine = sourceLine
+      closestElement = el
+    }
+  }
+
+  return closestElement
+}
+
+/**
+ * Create fixed-position highlight overlay at element location
+ * with boundary detection to prevent overflow and toolbar overlap
+ */
+function createHighlightOverlay(element: HTMLElement, container: HTMLElement): void {
+  try {
+    // Remove previous overlay
+    const previousOverlay = document.body.querySelector('.highlight-overlay')
+    if (previousOverlay) {
+      previousOverlay.remove()
+    }
+
+    const editorWrapper = container.closest('.rich-editor-wrapper')
+
+    // Create overlay at element position
+    const rect = element.getBoundingClientRect()
+    const overlay = document.createElement('div')
+    overlay.className = 'highlight-overlay animation-locate-highlight'
+    overlay.style.position = 'fixed'
+    overlay.style.left = `${rect.left}px`
+    overlay.style.top = `${rect.top}px`
+    overlay.style.width = `${rect.width}px`
+    overlay.style.height = `${rect.height}px`
+    overlay.style.pointerEvents = 'none'
+    overlay.style.zIndex = '9999'
+    overlay.style.borderRadius = '4px'
+
+    document.body.appendChild(overlay)
+
+    // Update overlay position and visibility on scroll
+    const updatePosition = () => {
+      const newRect = element.getBoundingClientRect()
+      const newContainerRect = container.getBoundingClientRect()
+
+      // Update position
+      overlay.style.left = `${newRect.left}px`
+      overlay.style.top = `${newRect.top}px`
+      overlay.style.width = `${newRect.width}px`
+      overlay.style.height = `${newRect.height}px`
+
+      // Get current toolbar bottom (it might change)
+      const currentToolbar = editorWrapper?.querySelector('[class*="ToolbarWrapper"]')
+      const currentToolbarRect = currentToolbar?.getBoundingClientRect()
+      const currentToolbarBottom = currentToolbarRect ? currentToolbarRect.bottom : newContainerRect.top
+
+      // Check if overlay is within visible bounds
+      const overlayTop = newRect.top
+      const overlayBottom = newRect.bottom
+      const visibleTop = currentToolbarBottom // Don't overlap toolbar
+      const visibleBottom = newContainerRect.bottom
+
+      // Hide overlay if any part is outside the visible container area
+      if (overlayTop < visibleTop || overlayBottom > visibleBottom) {
+        overlay.style.opacity = '0'
+        overlay.style.visibility = 'hidden'
+      } else {
+        overlay.style.opacity = '1'
+        overlay.style.visibility = 'visible'
+      }
+    }
+
+    container.addEventListener('scroll', updatePosition)
+
+    // Auto-remove after animation
+    const handleAnimationEnd = () => {
+      overlay.remove()
+      container.removeEventListener('scroll', updatePosition)
+      overlay.removeEventListener('animationend', handleAnimationEnd)
+    }
+    overlay.addEventListener('animationend', handleAnimationEnd)
+  } catch (error) {
+    logger.error('Failed to create highlight overlay:', error as Error)
+  }
+}
+
+/**
+ * Scroll to element and show highlight after scroll completes
+ */
+function scrollAndHighlight(element: HTMLElement, container: HTMLElement): void {
+  element.scrollIntoView({ behavior: 'smooth', block: 'start', inline: 'nearest' })
+
+  let scrollTimeout: NodeJS.Timeout
+  const handleScroll = () => {
+    clearTimeout(scrollTimeout)
+    scrollTimeout = setTimeout(() => {
+      container.removeEventListener('scroll', handleScroll)
+      requestAnimationFrame(() => createHighlightOverlay(element, container))
+    }, 150)
+  }
+
+  container.addEventListener('scroll', handleScroll)
+
+  // Fallback: if element already in view (no scroll happens)
+  setTimeout(() => {
+    const initialScrollTop = container.scrollTop
+    setTimeout(() => {
+      if (Math.abs(container.scrollTop - initialScrollTop) < 1) {
+        container.removeEventListener('scroll', handleScroll)
+        clearTimeout(scrollTimeout)
+        requestAnimationFrame(() => createHighlightOverlay(element, container))
+      }
+    }, 200)
+  }, 50)
+}
+
 const RichEditor = ({
   ref,
   initialContent = '',
@@ -372,6 +523,22 @@ const RichEditor = ({
           scrollContainerRef.current.scrollTop = value
         }
       },
+      scrollToLine: (lineNumber: number, options?: { highlight?: boolean; lineContent?: string }) => {
+        if (!editor || !scrollContainerRef.current) return
+
+        try {
+          const element = findElementByLine(editor.view.dom, lineNumber, options?.lineContent)
+          if (!element) return
+
+          if (options?.highlight) {
+            scrollAndHighlight(element, scrollContainerRef.current)
+          } else {
+            element.scrollIntoView({ behavior: 'smooth', block: 'start', inline: 'nearest' })
+          }
+        } catch (error) {
+          logger.error('Failed in scrollToLine:', error as Error)
+        }
+      },
       // Dynamic command management
       registerCommand,
       registerToolbarCommand,

src/renderer/src/components/RichEditor/types.ts

@@ -111,6 +111,8 @@ export interface RichEditorRef {
   getScrollTop: () => number
   /** Set scrollTop of the editor scroll container */
   setScrollTop: (value: number) => void
+  /** Scroll to specific line number in markdown */
+  scrollToLine: (lineNumber: number, options?: { highlight?: boolean; lineContent?: string }) => void
   // Dynamic command management
   /** Register a new command/toolbar item */
   registerCommand: (cmd: Command) => void

src/renderer/src/components/RichEditor/useRichEditor.ts

@@ -2,6 +2,7 @@ import 'katex/dist/katex.min.css'
 
 import { TableKit } from '@cherrystudio/extension-table-plus'
 import { loggerService } from '@logger'
+import { MARKDOWN_SOURCE_LINE_ATTR } from '@renderer/components/RichEditor/constants'
 import type { FormattingState } from '@renderer/components/RichEditor/types'
 import { useCodeStyle } from '@renderer/context/CodeStyleProvider'
 import {
@@ -11,6 +12,7 @@ import {
   markdownToPreviewText
 } from '@renderer/utils/markdownConverter'
 import type { Editor } from '@tiptap/core'
+import { Extension } from '@tiptap/core'
 import { TaskItem, TaskList } from '@tiptap/extension-list'
 import { migrateMathStrings } from '@tiptap/extension-mathematics'
 import Mention from '@tiptap/extension-mention'
@@ -36,6 +38,31 @@ import { blobToArrayBuffer, compressImage, shouldCompressImage } from './helpers
 
 const logger = loggerService.withContext('useRichEditor')
 
+// Create extension to preserve data-source-line attribute
+const SourceLineAttribute = Extension.create({
+  name: 'sourceLineAttribute',
+  addGlobalAttributes() {
+    return [
+      {
+        types: ['paragraph', 'heading', 'blockquote', 'bulletList', 'orderedList', 'listItem', 'horizontalRule'],
+        attributes: {
+          dataSourceLine: {
+            default: null,
+            parseHTML: (element) => {
+              const value = element.getAttribute(MARKDOWN_SOURCE_LINE_ATTR)
+              return value
+            },
+            renderHTML: (attributes) => {
+              if (!attributes.dataSourceLine) return {}
+              return { [MARKDOWN_SOURCE_LINE_ATTR]: attributes.dataSourceLine }
+            }
+          }
+        }
+      }
+    ]
+  }
+})
+
 export interface UseRichEditorOptions {
   /** Initial markdown content */
   initialContent?: string
@@ -196,6 +223,7 @@ export const useRichEditor = (options: UseRichEditorOptions = {}): UseRichEditor
   // TipTap editor extensions
   const extensions = useMemo(
     () => [
+      SourceLineAttribute,
       StarterKit.configure({
         heading: {
           levels: [1, 2, 3, 4, 5, 6]

src/renderer/src/components/Tab/TabContainer.tsx

@@ -32,6 +32,7 @@ import {
   Sparkle,
   Sun,
   Terminal,
+  Video,
   X
 } from 'lucide-react'
 import { useCallback, useEffect, useMemo } from 'react'
@@ -106,6 +107,8 @@ const getTabIcon = (
       return <Settings size={14} />
     case 'code':
       return <Terminal size={14} />
+    case 'video':
+      return <Video size={14} />
     default:
       return null
   }

src/renderer/src/components/ToastPortal.tsx

@@ -23,7 +23,8 @@ export const ToastPortal = () => {
         timeout: 3000,
         classNames: {
           // This setting causes the 'hero-toast' class to be applied twice to the toast element. This is weird and I don't know why, but it works.
-          base: 'hero-toast'
+          // `w-auto` would not overwrite default style, which set the width to a fixed value and causes text overflow.
+          base: 'hero-toast w-auto! max-w-[50vw]'
         }
       }}
     />,

src/renderer/src/components/TopView/toast.ts

@@ -1,5 +1,6 @@
 import { addToast, closeAll, closeToast, getToastQueue, isToastClosing } from '@heroui/toast'
 import { RequireSome } from '@renderer/types'
+import { t } from 'i18next'
 
 type AddToastProps = Parameters<typeof addToast>[0]
 type ToastPropsColored = Omit<AddToastProps, 'color'>
@@ -54,7 +55,7 @@ export const loading = (args: RequireSome<AddToastProps, 'promise'>) => {
   if (args.timeout === undefined) {
     args.timeout = 1
   }
-  return addToast(args)
+  return addToast({ title: t('common.loading'), ...args })
 }
 
 export const getToastUtilities = () =>

src/renderer/src/components/app/Navbar.tsx

@@ -74,7 +74,7 @@ const NavbarContainer = styled.div<{ $isFullScreen: boolean }>`
 `
 
 const NavbarLeftContainer = styled.div`
-  min-width: var(--assistants-width);
+  min-width: ${isMac ? 'calc(var(--assistants-width) - 20px)' : 'var(--assistants-width)'};
   padding: 0 10px;
   display: flex;
   flex-direction: row;

src/renderer/src/components/app/Sidebar.tsx

@@ -26,7 +26,8 @@ import {
   Palette,
   Settings,
   Sparkle,
-  Sun
+  Sun,
+  Video
 } from 'lucide-react'
 import { FC } from 'react'
 import { useTranslation } from 'react-i18next'
@@ -139,7 +140,8 @@ const MainMenus: FC = () => {
     knowledge: <FileSearch size={18} className="icon" />,
     files: <Folder size={18} className="icon" />,
     notes: <NotepadText size={18} className="icon" />,
-    code_tools: <Code size={18} className="icon" />
+    code_tools: <Code size={18} className="icon" />,
+    video: <Video size={18} className="icon" />
   }
 
   const pathMap = {
@@ -151,7 +153,8 @@ const MainMenus: FC = () => {
     knowledge: '/knowledge',
     files: '/files',
     code_tools: '/code',
-    notes: '/notes'
+    notes: '/notes',
+    video: '/video'
   }
 
   return sidebarIcons.visible.map((icon) => {

src/renderer/src/config/models/reasoning.ts

@@ -14,7 +14,7 @@ import { GEMINI_FLASH_MODEL_REGEX } from './websearch'
 
 // Reasoning models
 export const REASONING_REGEX =
-  /^(o\d+(?:-[\w-]+)?|.*\b(?:reasoning|reasoner|thinking)\b.*|.*-[rR]\d+.*|.*\bqwq(?:-[\w-]+)?\b.*|.*\bhunyuan-t1(?:-[\w-]+)?\b.*|.*\bglm-zero-preview\b.*|.*\bgrok-(?:3-mini|4)(?:-[\w-]+)?\b.*)$/i
+  /^(?!.*-non-reasoning\b)(o\d+(?:-[\w-]+)?|.*\b(?:reasoning|reasoner|thinking)\b.*|.*-[rR]\d+.*|.*\bqwq(?:-[\w-]+)?\b.*|.*\bhunyuan-t1(?:-[\w-]+)?\b.*|.*\bglm-zero-preview\b.*|.*\bgrok-(?:3-mini|4|4-fast)(?:-[\w-]+)?\b.*)$/i
 
 // 模型类型到支持的reasoning_effort的映射表
 // TODO: refactor this. too many identical options
@@ -24,6 +24,7 @@ export const MODEL_SUPPORTED_REASONING_EFFORT: ReasoningEffortConfig = {
   gpt5: ['minimal', 'low', 'medium', 'high'] as const,
   gpt5_codex: ['low', 'medium', 'high'] as const,
   grok: ['low', 'high'] as const,
+  grok4_fast: ['auto'] as const,
   gemini: ['low', 'medium', 'high', 'auto'] as const,
   gemini_pro: ['low', 'medium', 'high', 'auto'] as const,
   qwen: ['low', 'medium', 'high'] as const,
@@ -43,6 +44,7 @@ export const MODEL_SUPPORTED_OPTIONS: ThinkingOptionConfig = {
   gpt5: [...MODEL_SUPPORTED_REASONING_EFFORT.gpt5] as const,
   gpt5_codex: MODEL_SUPPORTED_REASONING_EFFORT.gpt5_codex,
   grok: MODEL_SUPPORTED_REASONING_EFFORT.grok,
+  grok4_fast: ['off', ...MODEL_SUPPORTED_REASONING_EFFORT.grok4_fast] as const,
   gemini: ['off', ...MODEL_SUPPORTED_REASONING_EFFORT.gemini] as const,
   gemini_pro: MODEL_SUPPORTED_REASONING_EFFORT.gemini_pro,
   qwen: ['off', ...MODEL_SUPPORTED_REASONING_EFFORT.qwen] as const,
@@ -66,6 +68,8 @@ export const getThinkModelType = (model: Model): ThinkingModelType => {
     }
   } else if (isSupportedReasoningEffortOpenAIModel(model)) {
     thinkingModelType = 'o'
+  } else if (isGrok4FastReasoningModel(model)) {
+    thinkingModelType = 'grok4_fast'
   } else if (isSupportedThinkingTokenGeminiModel(model)) {
     if (GEMINI_FLASH_MODEL_REGEX.test(model.id)) {
       thinkingModelType = 'gemini'
@@ -142,19 +146,46 @@ export function isSupportedReasoningEffortGrokModel(model?: Model): boolean {
   }
 
   const modelId = getLowerBaseModelName(model.id)
+  const providerId = model.provider.toLowerCase()
   if (modelId.includes('grok-3-mini')) {
     return true
   }
 
+  if (providerId === 'openrouter' && modelId.includes('grok-4-fast')) {
+    return true
+  }
+
   return false
 }
 
+/**
+ * Checks if the model is Grok 4 Fast reasoning version
+ * Explicitly excludes non-reasoning variants (models with 'non-reasoning' in their ID)
+ *
+ * Note: XAI official uses different model IDs for reasoning vs non-reasoning
+ * Third-party providers like OpenRouter expose a single ID with reasoning parameters, while first-party providers require separate IDs. Only the OpenRouter variant supports toggling.
+ *
+ * @param model - The model to check
+ * @returns true if the model is a reasoning-enabled Grok 4 Fast model
+ */
+export function isGrok4FastReasoningModel(model?: Model): boolean {
+  if (!model) {
+    return false
+  }
+
+  const modelId = getLowerBaseModelName(model.id)
+  return modelId.includes('grok-4-fast') && !modelId.includes('non-reasoning')
+}
+
 export function isGrokReasoningModel(model?: Model): boolean {
   if (!model) {
     return false
   }
   const modelId = getLowerBaseModelName(model.id)
-  if (isSupportedReasoningEffortGrokModel(model) || modelId.includes('grok-4')) {
+  if (
+    isSupportedReasoningEffortGrokModel(model) ||
+    (modelId.includes('grok-4') && !modelId.includes('non-reasoning'))
+  ) {
     return true
   }
 
@@ -265,7 +296,11 @@ export function isQwenAlwaysThinkModel(model?: Model): boolean {
     return false
   }
   const modelId = getLowerBaseModelName(model.id, '/')
-  return modelId.startsWith('qwen3') && modelId.includes('thinking')
+  // 包括 qwen3 开头的 thinking 模型和 qwen3-vl 的 thinking 模型
+  return (
+    (modelId.startsWith('qwen3') && modelId.includes('thinking')) ||
+    (modelId.includes('qwen3-vl') && modelId.includes('thinking'))
+  )
 }
 
 // Doubao 支持思考模式的模型正则
@@ -329,7 +364,10 @@ export const isPerplexityReasoningModel = (model?: Model): boolean => {
   }
 
   const modelId = getLowerBaseModelName(model.id, '/')
-  return isSupportedReasoningEffortPerplexityModel(model) || modelId.includes('reasoning')
+  return (
+    isSupportedReasoningEffortPerplexityModel(model) ||
+    (modelId.includes('reasoning') && !modelId.includes('non-reasoning'))
+  )
 }
 
 export const isSupportedReasoningEffortPerplexityModel = (model: Model): boolean => {
@@ -443,6 +481,8 @@ export const THINKING_TOKEN_MAP: Record<string, { min: number; max: number }> =
   // qwen-plus-x 系列自 qwen-plus-2025-07-28 后模型最长思维链变为 81_920, qwen-plus 模型于 2025.9.16 同步变更
   'qwen3-235b-a22b-thinking-2507$': { min: 0, max: 81_920 },
   'qwen3-30b-a3b-thinking-2507$': { min: 0, max: 81_920 },
+  'qwen3-vl-235b-a22b-thinking$': { min: 0, max: 81_920 },
+  'qwen3-vl-30b-a3b-thinking$': { min: 0, max: 81_920 },
   'qwen-plus-2025-07-14$': { min: 0, max: 38_912 },
   'qwen-plus-2025-04-28$': { min: 0, max: 38_912 },
   'qwen3-1\\.7b$': { min: 0, max: 30_720 },

src/renderer/src/config/models/utils.ts

@@ -1,11 +1,12 @@
+import OpenAI from '@cherrystudio/openai'
+import { isEmbeddingModel, isRerankModel } from '@renderer/config/models/embedding'
 import { Model } from '@renderer/types'
 import { getLowerBaseModelName } from '@renderer/utils'
-import OpenAI from 'openai'
 
 import { WEB_SEARCH_PROMPT_FOR_OPENROUTER } from '../prompts'
 import { getWebSearchTools } from '../tools'
 import { isOpenAIReasoningModel } from './reasoning'
-import { isGenerateImageModel, isVisionModel } from './vision'
+import { isGenerateImageModel, isTextToImageModel, isVisionModel } from './vision'
 import { isOpenAIWebSearchChatCompletionOnlyModel } from './websearch'
 export const NOT_SUPPORTED_REGEX = /(?:^tts|whisper|speech)/i
 
@@ -246,3 +247,7 @@ export const isOpenAIOpenWeightModel = (model: Model) => {
 
 // zhipu 视觉推理模型用这组 special token 标记推理结果
 export const ZHIPU_RESULT_TOKENS = ['<|begin_of_box|>', '<|end_of_box|>'] as const
+
+export const agentModelFilter = (model: Model): boolean => {
+  return !isEmbeddingModel(model) && !isRerankModel(model) && !isTextToImageModel(model)
+}

src/renderer/src/config/models/video.ts

@@ -0,0 +1,149 @@
+import { SystemProviderId, Video } from '@renderer/types'
+
+// Hard-encoded for now. We may implement a function to filter video generation model from provider.models.
+export const videoModelsMap = {
+  openai: ['sora-2', 'sora-2-pro'] as const
+} as const satisfies Partial<Record<SystemProviderId, string[]>>
+
+// Mock data for testing
+export const mockVideos: Video[] = [
+  {
+    id: '1',
+    type: 'openai',
+    status: 'downloaded',
+    prompt: 'A beautiful sunset over the ocean with waves crashing',
+    thumbnail: 'https://picsum.photos/200/200?random=1',
+    fileId: 'file-001',
+    providerId: 'openai',
+    name: 'video-001',
+    metadata: {
+      id: 'video-001',
+      object: 'video',
+      created_at: Math.floor(Date.now() / 1000),
+      completed_at: Math.floor(Date.now() / 1000),
+      expires_at: null,
+      error: null,
+      model: 'sora-2',
+      progress: 100,
+      remixed_from_video_id: null,
+      seconds: '4',
+      size: '1280x720',
+      status: 'completed'
+    }
+  },
+  {
+    id: '2',
+    type: 'openai',
+    status: 'in_progress',
+    prompt: 'A cat playing with a ball of yarn in slow motion',
+    progress: 65,
+    providerId: 'openai',
+    name: 'video-002',
+    metadata: {
+      id: 'video-002',
+      object: 'video',
+      created_at: Math.floor(Date.now() / 1000),
+      completed_at: null,
+      expires_at: null,
+      error: null,
+      model: 'sora-2-pro',
+      progress: 65,
+      remixed_from_video_id: null,
+      seconds: '8',
+      size: '1792x1024',
+      status: 'in_progress'
+    }
+  },
+  {
+    id: '3',
+    type: 'openai',
+    status: 'queued',
+    prompt: 'Time-lapse of flowers blooming in a garden',
+    providerId: 'openai',
+    name: 'video-003',
+    metadata: {
+      id: 'video-003',
+      object: 'video',
+      created_at: Math.floor(Date.now() / 1000),
+      completed_at: null,
+      expires_at: null,
+      error: null,
+      model: 'sora-2',
+      progress: 0,
+      remixed_from_video_id: null,
+      seconds: '12',
+      size: '1280x720',
+      status: 'queued'
+    }
+  },
+  {
+    id: '4',
+    type: 'openai',
+    prompt: 'Birds flying in formation against blue sky',
+    status: 'downloading',
+    progress: 80,
+    thumbnail: 'https://picsum.photos/200/200?random=4',
+    providerId: 'openai',
+    name: 'video-004',
+    metadata: {
+      id: 'video-004',
+      object: 'video',
+      created_at: Math.floor(Date.now() / 1000),
+      completed_at: Math.floor(Date.now() / 1000),
+      expires_at: null,
+      error: null,
+      model: 'sora-2-pro',
+      progress: 100,
+      remixed_from_video_id: null,
+      seconds: '8',
+      size: '1792x1024',
+      status: 'completed'
+    }
+  },
+  {
+    id: '5',
+    type: 'openai',
+    status: 'failed',
+    error: { code: '400', message: 'Video generation failed' },
+    prompt: 'Mountain landscape with snow peaks and forest',
+    providerId: 'openai',
+    name: 'video-005',
+    metadata: {
+      id: 'video-005',
+      object: 'video',
+      created_at: Math.floor(Date.now() / 1000),
+      completed_at: Math.floor(Date.now() / 1000),
+      expires_at: null,
+      error: { code: '400', message: 'Video generation failed' },
+      model: 'sora-2',
+      progress: 0,
+      remixed_from_video_id: null,
+      seconds: '4',
+      size: '1280x720',
+      status: 'failed'
+    }
+  },
+  {
+    id: '6',
+    type: 'openai',
+    status: 'completed',
+    thumbnail: 'https://picsum.photos/200/200?random=6',
+    prompt: 'City street at night with neon lights reflecting on wet pavement',
+    providerId: 'openai',
+    name: 'video-006',
+    metadata: {
+      id: 'video-006',
+      object: 'video',
+      created_at: Math.floor(Date.now() / 1000),
+      completed_at: Math.floor(Date.now() / 1000),
+      expires_at: null,
+      error: null,
+      model: 'sora-2-pro',
+      progress: 100,
+      remixed_from_video_id: null,
+      seconds: '12',
+      size: '1024x1792',
+      status: 'completed'
+    }
+  }
+]

src/renderer/src/config/models/vision.ts

@@ -24,7 +24,7 @@ const visionAllowedModels = [
   'qwen2.5-vl',
   'qwen3-vl',
   'qwen2.5-omni',
-  'qwen3-omni',
+  'qwen3-omni(?:-[\\w-]+)?',
   'qvq',
   'internvl2',
   'grok-vision-beta',
@@ -82,14 +82,14 @@ export const IMAGE_ENHANCEMENT_MODELS = [
   'grok-2-image(?:-[\\w-]+)?',
   'qwen-image-edit',
   'gpt-image-1',
-  'gemini-2.5-flash-image-preview',
+  'gemini-2.5-flash-image',
   'gemini-2.0-flash-preview-image-generation'
 ]
 
 const IMAGE_ENHANCEMENT_MODELS_REGEX = new RegExp(IMAGE_ENHANCEMENT_MODELS.join('|'), 'i')
 
 // Models that should auto-enable image generation button when selected
-export const AUTO_ENABLE_IMAGE_MODELS = ['gemini-2.5-flash-image-preview', ...DEDICATED_IMAGE_MODELS]
+export const AUTO_ENABLE_IMAGE_MODELS = ['gemini-2.5-flash-image', ...DEDICATED_IMAGE_MODELS]
 
 export const OPENAI_TOOL_USE_IMAGE_GENERATION_MODELS = [
   'o3',
@@ -107,7 +107,7 @@ export const GENERATE_IMAGE_MODELS = [
   'gemini-2.0-flash-exp',
   'gemini-2.0-flash-exp-image-generation',
   'gemini-2.0-flash-preview-image-generation',
-  'gemini-2.5-flash-image-preview',
+  'gemini-2.5-flash-image',
   ...DEDICATED_IMAGE_MODELS
 ]
 

src/renderer/src/config/ocr.ts

@@ -1,6 +1,7 @@
 import {
   BuiltinOcrProvider,
   BuiltinOcrProviderId,
+  OcrOvProvider,
   OcrPpocrProvider,
   OcrProviderCapability,
   OcrSystemProvider,
@@ -50,10 +51,23 @@ const ppocrOcr: OcrPpocrProvider = {
   }
 } as const
 
+const ovOcr: OcrOvProvider = {
+  id: 'ovocr',
+  name: 'Intel OV(NPU) OCR',
+  config: {
+    langs: isWin ? ['en-us', 'zh-cn'] : undefined
+  },
+  capabilities: {
+    image: true
+    // pdf: true
+  }
+} as const satisfies OcrOvProvider
+
 export const BUILTIN_OCR_PROVIDERS_MAP = {
   tesseract,
   system: systemOcr,
-  paddleocr: ppocrOcr
+  paddleocr: ppocrOcr,
+  ovocr: ovOcr
 } as const satisfies Record<BuiltinOcrProviderId, BuiltinOcrProvider>
 
 export const BUILTIN_OCR_PROVIDERS: BuiltinOcrProvider[] = Object.values(BUILTIN_OCR_PROVIDERS_MAP)

src/renderer/src/config/providers.ts

@@ -58,7 +58,6 @@ import ZhipuProviderLogo from '@renderer/assets/images/providers/zhipu.png'
 import {
   AtLeast,
   isSystemProvider,
-  Model,
   OpenAIServiceTiers,
   Provider,
   ProviderType,
@@ -88,6 +87,7 @@ export const SYSTEM_PROVIDERS_CONFIG: Record<SystemProviderId, SystemProvider> =
     type: 'openai',
     apiKey: '',
     apiHost: 'https://open.cherryin.net',
+    anthropicApiHost: 'https://open.cherryin.net',
     models: [],
     isSystem: true,
     enabled: true
@@ -109,7 +109,6 @@ export const SYSTEM_PROVIDERS_CONFIG: Record<SystemProviderId, SystemProvider> =
     apiKey: '',
     apiHost: 'https://aihubmix.com',
     anthropicApiHost: 'https://aihubmix.com/anthropic',
-    isAnthropicModel: (m: Model) => m.id.includes('claude'),
     models: SYSTEM_MODELS.aihubmix,
     isSystem: true,
     enabled: false
@@ -289,7 +288,7 @@ export const SYSTEM_PROVIDERS_CONFIG: Record<SystemProviderId, SystemProvider> =
   'new-api': {
     id: 'new-api',
     name: 'New API',
-    type: 'openai',
+    type: 'new-api',
     apiKey: '',
     apiHost: 'http://localhost:3000',
     anthropicApiHost: 'http://localhost:3000',
@@ -1432,5 +1431,5 @@ export const isGeminiWebSearchProvider = (provider: Provider) => {
 }
 
 export const isNewApiProvider = (provider: Provider) => {
-  return ['new-api', 'cherryin'].includes(provider.id)
+  return ['new-api', 'cherryin'].includes(provider.id) || provider.type === 'new-api'
 }

src/renderer/src/config/tools.ts

@@ -1,5 +1,5 @@
+import { ChatCompletionTool } from '@cherrystudio/openai/resources'
 import { Model } from '@renderer/types'
-import { ChatCompletionTool } from 'openai/resources'
 
 import { WEB_SEARCH_PROMPT_FOR_ZHIPU } from './prompts'
 

src/renderer/src/hooks/agents/types.ts

@@ -0,0 +1,4 @@
+export type UpdateAgentBaseOptions = {
+  /** Whether to show success toast after updating. Defaults to true. */
+  showSuccessToast?: boolean
+}

src/renderer/src/hooks/agents/useActiveAgent.ts

@@ -0,0 +1,8 @@
+import { useRuntime } from '../useRuntime'
+import { useAgent } from './useAgent'
+
+export const useActiveAgent = () => {
+  const { chat } = useRuntime()
+  const { activeAgentId } = chat
+  return useAgent(activeAgentId)
+}

src/renderer/src/hooks/agents/useActiveSession.ts

@@ -0,0 +1,9 @@
+import { useRuntime } from '../useRuntime'
+import { useSession } from './useSession'
+
+export const useActiveSession = () => {
+  const { chat } = useRuntime()
+  const { activeSessionIdMap, activeAgentId } = chat
+  const activeSessionId = activeAgentId ? activeSessionIdMap[activeAgentId] : null
+  return useSession(activeAgentId, activeSessionId)
+}

src/renderer/src/hooks/agents/useAgent.ts

@@ -1,18 +1,28 @@
 import { useCallback } from 'react'
+import { useTranslation } from 'react-i18next'
 import useSWR from 'swr'
 
+import { useApiServer } from '../useApiServer'
 import { useAgentClient } from './useAgentClient'
 
 export const useAgent = (id: string | null) => {
+  const { t } = useTranslation()
   const client = useAgentClient()
   const key = id ? client.agentPaths.withId(id) : null
+  const { apiServerConfig, apiServerRunning } = useApiServer()
   const fetcher = useCallback(async () => {
-    if (!id || id === 'fake') {
-      return null
+    if (!id) {
+      throw new Error(t('agent.get.error.null_id'))
+    }
+    if (!apiServerConfig.enabled) {
+      throw new Error(t('apiServer.messages.notEnabled'))
+    }
+    if (!apiServerRunning) {
+      throw new Error(t('agent.server.error.not_running'))
     }
     const result = await client.getAgent(id)
     return result
-  }, [client, id])
+  }, [apiServerConfig.enabled, apiServerRunning, client, id, t])
   const { data, error, isLoading } = useSWR(key, id ? fetcher : null)
 
   return {

src/renderer/src/hooks/agents/useAgentSessionInitializer.ts

@@ -17,7 +17,7 @@ export const useAgentSessionInitializer = () => {
   const dispatch = useAppDispatch()
   const client = useAgentClient()
   const { chat } = useRuntime()
-  const { activeAgentId, activeSessionId } = chat
+  const { activeAgentId, activeSessionIdMap } = chat
 
   /**
    * Initialize session for the given agent by loading its sessions
@@ -25,11 +25,11 @@ export const useAgentSessionInitializer = () => {
    */
   const initializeAgentSession = useCallback(
     async (agentId: string) => {
-      if (!agentId || agentId === 'fake') return
+      if (!agentId) return
 
       try {
         // Check if this agent already has an active session
-        const currentSessionId = activeSessionId[agentId]
+        const currentSessionId = activeSessionIdMap[agentId]
         if (currentSessionId) {
           // Session already exists, just switch to session view
           dispatch(setActiveTopicOrSessionAction('session'))
@@ -58,21 +58,21 @@ export const useAgentSessionInitializer = () => {
         dispatch(setActiveTopicOrSessionAction('session'))
       }
     },
-    [client, dispatch, activeSessionId]
+    [client, dispatch, activeSessionIdMap]
   )
 
   /**
    * Auto-initialize when activeAgentId changes
    */
   useEffect(() => {
-    if (activeAgentId && activeAgentId !== 'fake') {
+    if (activeAgentId) {
       // Check if we need to initialize this agent's session
-      const hasActiveSession = activeSessionId[activeAgentId]
+      const hasActiveSession = activeSessionIdMap[activeAgentId]
       if (!hasActiveSession) {
         initializeAgentSession(activeAgentId)
       }
     }
-  }, [activeAgentId, activeSessionId, initializeAgentSession])
+  }, [activeAgentId, activeSessionIdMap, initializeAgentSession])
 
   return {
     initializeAgentSession

src/renderer/src/hooks/agents/useAgents.ts

@@ -6,6 +6,7 @@ import { useCallback } from 'react'
 import { useTranslation } from 'react-i18next'
 import useSWR from 'swr'
 
+import { useApiServer } from '../useApiServer'
 import { useRuntime } from '../useRuntime'
 import { useAgentClient } from './useAgentClient'
 
@@ -23,11 +24,19 @@ export const useAgents = () => {
   const { t } = useTranslation()
   const client = useAgentClient()
   const key = client.agentPaths.base
+  const { apiServerConfig, apiServerRunning } = useApiServer()
   const fetcher = useCallback(async () => {
+    // API server will start on startup if enabled OR there are agents
+    if (!apiServerConfig.enabled && !apiServerRunning) {
+      throw new Error(t('apiServer.messages.notEnabled'))
+    }
+    if (!apiServerRunning) {
+      throw new Error(t('agent.server.error.not_running'))
+    }
     const result = await client.listAgents()
     // NOTE: We only use the array for now. useUpdateAgent depends on this behavior.
     return result.data
-  }, [client])
+  }, [apiServerConfig.enabled, apiServerRunning, client, t])
   const { data, error, isLoading, mutate } = useSWR(key, fetcher)
   const { chat } = useRuntime()
   const { activeAgentId } = chat

src/renderer/src/hooks/agents/useSession.ts

@@ -1,21 +1,24 @@
 import { useAppDispatch } from '@renderer/store'
 import { loadTopicMessagesThunk } from '@renderer/store/thunk/messageThunk'
-import { UpdateSessionForm } from '@renderer/types'
 import { buildAgentSessionTopicId } from '@renderer/utils/agentSession'
-import { useCallback, useEffect, useMemo } from 'react'
+import { useEffect, useMemo } from 'react'
 import { useTranslation } from 'react-i18next'
 import useSWR from 'swr'
 
 import { useAgentClient } from './useAgentClient'
+import { useUpdateSession } from './useUpdateSession'
 
-export const useSession = (agentId: string, sessionId: string) => {
+export const useSession = (agentId: string | null, sessionId: string | null) => {
   const { t } = useTranslation()
   const client = useAgentClient()
-  const key = client.getSessionPaths(agentId).withId(sessionId)
+  const key = agentId && sessionId ? client.getSessionPaths(agentId).withId(sessionId) : null
   const dispatch = useAppDispatch()
-  const sessionTopicId = useMemo(() => buildAgentSessionTopicId(sessionId), [sessionId])
+  const sessionTopicId = useMemo(() => (sessionId ? buildAgentSessionTopicId(sessionId) : null), [sessionId])
+  const { updateSession } = useUpdateSession(agentId)
 
   const fetcher = async () => {
+    if (!agentId) throw new Error(t('agent.get.error.null_id'))
+    if (!sessionId) throw new Error(t('agent.session.get.error.null_id'))
     const data = await client.getSession(agentId, sessionId)
     return data
   }
@@ -24,26 +27,13 @@ export const useSession = (agentId: string, sessionId: string) => {
   // Use loadTopicMessagesThunk to load messages (with caching mechanism)
   // This ensures messages are preserved when switching between sessions/tabs
   useEffect(() => {
-    if (sessionId) {
+    if (sessionTopicId) {
       // loadTopicMessagesThunk will check if messages already exist in Redux
       // and skip loading if they do (unless forceReload is true)
       dispatch(loadTopicMessagesThunk(sessionTopicId))
     }
   }, [dispatch, sessionId, sessionTopicId])
 
-  const updateSession = useCallback(
-    async (form: UpdateSessionForm) => {
-      if (!agentId) return
-      try {
-        const result = await client.updateSession(agentId, form)
-        mutate(result)
-      } catch (error) {
-        window.toast.error(t('agent.session.update.error.failed'))
-      }
-    },
-    [agentId, client, mutate, t]
-  )
-
   return {
     session: data,
     error,

src/renderer/src/hooks/agents/useSessions.ts

@@ -1,4 +1,4 @@
-import { CreateSessionForm } from '@renderer/types'
+import { CreateAgentSessionResponse, CreateSessionForm, GetAgentSessionResponse } from '@renderer/types'
 import { formatErrorMessageWithPrefix } from '@renderer/utils/error'
 import { useCallback } from 'react'
 import { useTranslation } from 'react-i18next'
@@ -6,46 +6,50 @@ import useSWR from 'swr'
 
 import { useAgentClient } from './useAgentClient'
 
-export const useSessions = (agentId: string) => {
+export const useSessions = (agentId: string | null) => {
   const { t } = useTranslation()
   const client = useAgentClient()
-  const key = client.getSessionPaths(agentId).base
+  const key = agentId ? client.getSessionPaths(agentId).base : null
 
   const fetcher = async () => {
+    if (!agentId) throw new Error('No active agent.')
     const data = await client.listSessions(agentId)
     return data.data
   }
   const { data, error, isLoading, mutate } = useSWR(key, fetcher)
 
   const createSession = useCallback(
-    async (form: CreateSessionForm) => {
+    async (form: CreateSessionForm): Promise<CreateAgentSessionResponse | null> => {
+      if (!agentId) return null
       try {
         const result = await client.createSession(agentId, form)
-        await mutate((prev) => [...(prev ?? []), result], { revalidate: false })
+        await mutate((prev) => [result, ...(prev ?? [])], { revalidate: false })
         return result
       } catch (error) {
         window.toast.error(formatErrorMessageWithPrefix(error, t('agent.session.create.error.failed')))
-        return undefined
+        return null
       }
     },
     [agentId, client, mutate, t]
   )
 
-  // TODO: including messages field
   const getSession = useCallback(
-    async (id: string) => {
+    async (id: string): Promise<GetAgentSessionResponse | null> => {
+      if (!agentId) return null
       try {
         const result = await client.getSession(agentId, id)
         mutate((prev) => prev?.map((session) => (session.id === result.id ? result : session)))
+        return result
       } catch (error) {
         window.toast.error(formatErrorMessageWithPrefix(error, t('agent.session.get.error.failed')))
+        return null
       }
     },
     [agentId, client, mutate, t]
   )
 
   const deleteSession = useCallback(
-    async (id: string) => {
+    async (id: string): Promise<boolean> => {
       if (!agentId) return false
       try {
         await client.deleteSession(agentId, id)

src/renderer/src/hooks/agents/useUpdateAgent.ts

@@ -4,20 +4,16 @@ import { useCallback } from 'react'
 import { useTranslation } from 'react-i18next'
 import { mutate } from 'swr'
 
+import { UpdateAgentBaseOptions } from './types'
 import { useAgentClient } from './useAgentClient'
 
-export type UpdateAgentOptions = {
-  /** Whether to show success toast after updating. Defaults to true. */
-  showSuccessToast?: boolean
-}
-
 export const useUpdateAgent = () => {
   const { t } = useTranslation()
   const client = useAgentClient()
   const listKey = client.agentPaths.base
 
   const updateAgent = useCallback(
-    async (form: UpdateAgentForm, options?: UpdateAgentOptions) => {
+    async (form: UpdateAgentForm, options?: UpdateAgentBaseOptions) => {
       try {
         const itemKey = client.agentPaths.withId(form.id)
         // may change to optimistic update
@@ -35,7 +31,7 @@ export const useUpdateAgent = () => {
   )
 
   const updateModel = useCallback(
-    async (agentId: string, modelId: string, options?: UpdateAgentOptions) => {
+    async (agentId: string, modelId: string, options?: UpdateAgentBaseOptions) => {
       updateAgent({ id: agentId, model: modelId }, options)
     },
     [updateAgent]

src/renderer/src/hooks/agents/useUpdateSession.ts

@@ -1,19 +1,21 @@
 import { ListAgentSessionsResponse, UpdateSessionForm } from '@renderer/types'
-import { formatErrorMessageWithPrefix } from '@renderer/utils/error'
+import { getErrorMessage } from '@renderer/utils/error'
 import { useCallback } from 'react'
 import { useTranslation } from 'react-i18next'
 import { mutate } from 'swr'
 
+import { UpdateAgentBaseOptions } from './types'
 import { useAgentClient } from './useAgentClient'
 
-export const useUpdateSession = (agentId: string) => {
+export const useUpdateSession = (agentId: string | null) => {
   const { t } = useTranslation()
   const client = useAgentClient()
-  const paths = client.getSessionPaths(agentId)
-  const listKey = paths.base
 
   const updateSession = useCallback(
-    async (form: UpdateSessionForm) => {
+    async (form: UpdateSessionForm, options?: UpdateAgentBaseOptions) => {
+      if (!agentId) return
+      const paths = client.getSessionPaths(agentId)
+      const listKey = paths.base
       const sessionId = form.id
       try {
         const itemKey = paths.withId(sessionId)
@@ -24,13 +26,29 @@ export const useUpdateSession = (agentId: string) => {
           (prev) => prev?.map((session) => (session.id === result.id ? result : session)) ?? []
         )
         mutate(itemKey, result)
-        window.toast.success(t('common.update_success'))
+        if (options?.showSuccessToast ?? true) {
+          window.toast.success(t('common.update_success'))
+        }
       } catch (error) {
-        window.toast.error(formatErrorMessageWithPrefix(error, t('agent.update.error.failed')))
+        window.toast.error({ title: t('agent.session.update.error.failed'), description: getErrorMessage(error) })
       }
     },
-    [agentId, client, listKey, paths, t]
+    [agentId, client, t]
   )
 
-  return updateSession
+  const updateModel = useCallback(
+    async (sessionId: string, modelId: string, options?: UpdateAgentBaseOptions) => {
+      if (!agentId) return
+      return updateSession(
+        {
+          id: sessionId,
+          model: modelId
+        },
+        options
+      )
+    },
+    [agentId, updateSession]
+  )
+
+  return { updateSession, updateModel }
 }

src/renderer/src/hooks/useApiServer.ts

@@ -0,0 +1,112 @@
+import { loggerService } from '@logger'
+import { useAppDispatch, useAppSelector } from '@renderer/store'
+import { setApiServerEnabled as setApiServerEnabledAction } from '@renderer/store/settings'
+import { useCallback, useEffect, useState } from 'react'
+import { useTranslation } from 'react-i18next'
+
+const logger = loggerService.withContext('useApiServer')
+
+export const useApiServer = () => {
+  const { t } = useTranslation()
+  // FIXME: We currently store two copies of the config data in both the renderer and the main processes,
+  // which carries the risk of data inconsistency. This should be modified so that the main process stores
+  // the data, and the renderer retrieves it.
+  const apiServerConfig = useAppSelector((state) => state.settings.apiServer)
+  const dispatch = useAppDispatch()
+
+  // Optimistic initial state.
+  const [apiServerRunning, setApiServerRunning] = useState(apiServerConfig.enabled)
+  const [apiServerLoading, setApiServerLoading] = useState(true)
+
+  const setApiServerEnabled = useCallback(
+    (enabled: boolean) => {
+      dispatch(setApiServerEnabledAction(enabled))
+    },
+    [dispatch]
+  )
+
+  // API Server functions
+  const checkApiServerStatus = useCallback(async () => {
+    setApiServerLoading(true)
+    try {
+      const status = await window.api.apiServer.getStatus()
+      setApiServerRunning(status.running)
+    } catch (error: any) {
+      logger.error('Failed to check API server status:', error)
+    } finally {
+      setApiServerLoading(false)
+    }
+  }, [])
+
+  const startApiServer = useCallback(async () => {
+    if (apiServerLoading) return
+
+    setApiServerLoading(true)
+    try {
+      const result = await window.api.apiServer.start()
+      if (result.success) {
+        setApiServerRunning(true)
+        window.toast.success(t('apiServer.messages.startSuccess'))
+      } else {
+        window.toast.error(t('apiServer.messages.startError') + result.error)
+      }
+    } catch (error: any) {
+      window.toast.error(t('apiServer.messages.startError') + (error.message || error))
+    } finally {
+      setApiServerLoading(false)
+    }
+  }, [apiServerLoading, t])
+
+  const stopApiServer = useCallback(async () => {
+    if (apiServerLoading) return
+
+    setApiServerLoading(true)
+    try {
+      const result = await window.api.apiServer.stop()
+      if (result.success) {
+        setApiServerRunning(false)
+        window.toast.success(t('apiServer.messages.stopSuccess'))
+      } else {
+        window.toast.error(t('apiServer.messages.stopError') + result.error)
+      }
+    } catch (error: any) {
+      window.toast.error(t('apiServer.messages.stopError') + (error.message || error))
+    } finally {
+      setApiServerLoading(false)
+    }
+  }, [apiServerLoading, t])
+
+  const restartApiServer = useCallback(async () => {
+    if (apiServerLoading) return
+
+    setApiServerLoading(true)
+    try {
+      const result = await window.api.apiServer.restart()
+      if (result.success) {
+        await checkApiServerStatus()
+        window.toast.success(t('apiServer.messages.restartSuccess'))
+      } else {
+        window.toast.error(t('apiServer.messages.restartError') + result.error)
+      }
+    } catch (error) {
+      window.toast.error(t('apiServer.messages.restartFailed') + (error as Error).message)
+    } finally {
+      setApiServerLoading(false)
+    }
+  }, [apiServerLoading, checkApiServerStatus, t])
+
+  useEffect(() => {
+    checkApiServerStatus()
+  }, [checkApiServerStatus])
+
+  return {
+    apiServerConfig,
+    apiServerRunning,
+    apiServerLoading,
+    startApiServer,
+    stopApiServer,
+    restartApiServer,
+    checkApiServerStatus,
+    setApiServerEnabled
+  }
+}

src/renderer/src/hooks/useOcrProvider.tsx

@@ -1,4 +1,5 @@
 import { loggerService } from '@logger'
+import IntelLogo from '@renderer/assets/images/providers/intel.png'
 import PaddleocrLogo from '@renderer/assets/images/providers/paddleocr.png'
 import TesseractLogo from '@renderer/assets/images/providers/Tesseract.js.png'
 import { BUILTIN_OCR_PROVIDERS_MAP, DEFAULT_OCR_PROVIDER } from '@renderer/config/ocr'
@@ -83,6 +84,8 @@ export const useOcrProviders = () => {
           return <MonitorIcon size={size} />
         case 'paddleocr':
           return <Avatar size={size} src={PaddleocrLogo} />
+        case 'ovocr':
+          return <Avatar size={size} src={IntelLogo} />
       }
     }
     return <FileQuestionMarkIcon size={size} />

src/renderer/src/hooks/usePending.ts

@@ -0,0 +1,17 @@
+import { useAppDispatch } from '@renderer/store'
+import { setPendingAction } from '@renderer/store/runtime'
+import { useCallback } from 'react'
+
+import { useRuntime } from './useRuntime'
+
+export const usePending = () => {
+  const { pendingMap } = useRuntime()
+  const dispatch = useAppDispatch()
+  const setPending = useCallback(
+    (id: string, value: boolean | undefined) => {
+      dispatch(setPendingAction({ id, value }))
+    },
+    [dispatch]
+  )
+  return { pendingMap, setPending }
+}

src/renderer/src/hooks/video/useAddOpenAIVideo.ts

@@ -0,0 +1,65 @@
+import OpenAI from '@cherrystudio/openai'
+import { useCallback } from 'react'
+
+import { useProviderVideos } from './useProviderVideos'
+
+export const useAddOpenAIVideo = (providerId: string) => {
+  const { addVideo } = useProviderVideos(providerId)
+
+  const addOpenAIVideo = useCallback(
+    (video: OpenAI.Videos.Video, prompt: string) => {
+      switch (video.status) {
+        case 'queued':
+          addVideo({
+            id: video.id,
+            name: video.id,
+            providerId,
+            status: video.status,
+            type: 'openai',
+            metadata: video,
+            prompt
+          })
+          break
+        case 'in_progress':
+          addVideo({
+            id: video.id,
+            name: video.id,
+            providerId,
+            status: 'in_progress',
+            type: 'openai',
+            progress: video.progress,
+            metadata: video,
+            prompt
+          })
+          break
+        case 'completed':
+          addVideo({
+            id: video.id,
+            name: video.id,
+            providerId,
+            status: 'completed',
+            type: 'openai',
+            metadata: video,
+            prompt,
+            thumbnail: null
+          })
+          break
+        case 'failed':
+          addVideo({
+            id: video.id,
+            name: video.id,
+            providerId,
+            status: 'failed',
+            type: 'openai',
+            error: video.error,
+            metadata: video,
+            prompt
+          })
+          break
+      }
+    },
+    [addVideo, providerId]
+  )
+
+  return addOpenAIVideo
+}

src/renderer/src/hooks/video/useOpenAIVideo.ts

@@ -0,0 +1,47 @@
+import { retrieveVideo } from '@renderer/services/ApiService'
+import useSWR, { SWRConfiguration, useSWRConfig } from 'swr'
+
+import { useProvider } from '../useProvider'
+import { useVideo } from './useVideo'
+
+export const useOpenAIVideo = (providerId: string, id: string) => {
+  const { provider } = useProvider(providerId)
+  const fetcher = async () => {
+    switch (provider.type) {
+      case 'openai-response':
+        return retrieveVideo({
+          type: 'openai',
+          videoId: id,
+          provider
+        })
+
+      default:
+        throw new Error(`Unsupported provider type: ${provider.type}`)
+    }
+  }
+  const video = useVideo(providerId, id)
+  let options: SWRConfiguration = {}
+  switch (video?.status) {
+    case 'queued':
+    case 'in_progress':
+      options = {
+        refreshInterval: 3000
+      }
+      break
+    default:
+      options = {
+        revalidateOnFocus: false,
+        revalidateOnMount: true
+      }
+  }
+  const { data, isLoading, error } = useSWR(`video/openai/${id}`, fetcher, options)
+  const { mutate } = useSWRConfig()
+  const revalidate = () => mutate(`video/openai/${id}`)
+
+  return {
+    video: data,
+    isLoading,
+    error,
+    revalidate
+  }
+}

src/renderer/src/hooks/video/useProviderVideos.ts

@@ -0,0 +1,174 @@
+import { loggerService } from '@logger'
+import { retrieveVideo } from '@renderer/services/ApiService'
+import { getProviderById } from '@renderer/services/ProviderService'
+import { useAppDispatch, useAppSelector } from '@renderer/store'
+import { addVideoAction, setVideoAction, setVideosAction, updateVideoAction } from '@renderer/store/video'
+import { Video } from '@renderer/types/video'
+import { getErrorMessage } from '@renderer/utils'
+import { useCallback, useEffect, useRef } from 'react'
+import { useTranslation } from 'react-i18next'
+import useSWR from 'swr'
+
+import { useVideos } from './useVideos'
+import { useVideoThumbnail } from './useVideoThumbnail'
+
+const logger = loggerService.withContext('useVideo')
+
+export const useProviderVideos = (providerId: string) => {
+  const { removeVideo } = useVideos()
+  const videos = useAppSelector((state) => state.video.videoMap[providerId])
+  const videosRef = useRef(videos)
+  const dispatch = useAppDispatch()
+  const { t } = useTranslation()
+
+  useEffect(() => {
+    videosRef.current = videos
+  }, [videos])
+
+  const getVideo = useCallback(
+    (id: string) => {
+      return videos?.find((v) => v.id === id)
+    },
+    [videos]
+  )
+
+  const addVideo = useCallback(
+    (video: Video) => {
+      if (videos && videos.every((v) => v.id !== video.id)) {
+        dispatch(addVideoAction({ providerId, video }))
+      }
+    },
+    [dispatch, providerId, videos]
+  )
+
+  const updateVideo = useCallback(
+    (update: Partial<Omit<Video, 'status'>> & { id: string }) => {
+      dispatch(updateVideoAction({ providerId, update }))
+    },
+    [dispatch, providerId]
+  )
+
+  const setVideo = useCallback(
+    (video: Video) => {
+      dispatch(setVideoAction({ providerId, video }))
+    },
+    [dispatch, providerId]
+  )
+
+  const setVideos = useCallback(
+    (newVideos: Video[]) => {
+      dispatch(setVideosAction({ providerId, videos: newVideos }))
+    },
+    [dispatch, providerId]
+  )
+
+  const removeProviderVideo = useCallback(
+    (videoId: string) => {
+      removeVideo(videoId, providerId)
+    },
+    [providerId, removeVideo]
+  )
+
+  useEffect(() => {
+    if (!videos) {
+      setVideos([])
+    }
+  }, [setVideos, videos])
+
+  // update videos from api
+  // NOTE: This provider should support openai videos endpoint. No runtime check here.
+  const provider = getProviderById(providerId)
+  const fetcher = async () => {
+    if (!videos || !provider) return []
+    if (provider.type === 'openai-response') {
+      const openaiVideos = videos
+        .filter((v) => v.type === 'openai')
+        .filter((v) => v.status === 'queued' || v.status === 'in_progress')
+      const jobs = openaiVideos.map((v) => retrieveVideo({ type: 'openai', videoId: v.id, provider }))
+      const result = await Promise.allSettled(jobs)
+      return result.filter((p) => p.status === 'fulfilled').map((p) => p.value)
+    } else {
+      throw new Error(`Provider type ${provider.type} is not supported for video status polling`)
+    }
+  }
+  const { data, error } = useSWR('video/openai/videos', fetcher, { refreshInterval: 3000 })
+  const { retrieveThumbnail } = useVideoThumbnail()
+  useEffect(() => {
+    if (error) {
+      logger.error('Failed to fetch video status updates', error)
+      return
+    }
+    if (!provider) {
+      logger.warn(`Provider ${providerId} not found.`)
+      return
+    }
+    const videos = videosRef.current
+
+    if (!data || !videos) return
+    data.forEach((v) => {
+      const retrievedVideo = v.video
+      const storeVideo = videos.find((v) => v.id === retrievedVideo.id)
+      if (!storeVideo) {
+        logger.warn(`Try to update video ${retrievedVideo.id}, but it's not in the store.`)
+        return
+      }
+      switch (retrievedVideo.status) {
+        case 'in_progress':
+          if (storeVideo.status === 'queued' || storeVideo.status === 'in_progress') {
+            setVideo({
+              ...storeVideo,
+              status: 'in_progress',
+              progress: retrievedVideo.progress,
+              metadata: retrievedVideo
+            })
+          }
+          break
+        case 'completed': {
+          if (storeVideo.status === 'in_progress' || storeVideo.status === 'queued') {
+            const newVideo = { ...storeVideo, status: 'completed', thumbnail: null, metadata: retrievedVideo } as const
+            setVideo(newVideo)
+            // Try to get thumbnail
+            retrieveThumbnail(newVideo)
+              .then((thumbnail) => {
+                const latestVideo = videosRef.current?.find((v) => v.id === newVideo.id)
+                if (
+                  thumbnail !== null &&
+                  latestVideo &&
+                  latestVideo.status !== 'queued' &&
+                  latestVideo.status !== 'in_progress' &&
+                  latestVideo.status !== 'failed'
+                ) {
+                  setVideo({
+                    ...latestVideo,
+                    thumbnail
+                  })
+                }
+              })
+              .catch((e) => {
+                logger.error('Failed to get thumbnail', e as Error)
+                window.toast.error({ title: t('video.thumbnail.error.get'), description: getErrorMessage(e) })
+              })
+          }
+          break
+        }
+        case 'failed':
+          setVideo({
+            ...storeVideo,
+            status: 'failed',
+            error: retrievedVideo.error,
+            metadata: retrievedVideo
+          })
+      }
+    })
+  }, [data, error, provider, providerId, retrieveThumbnail, setVideo, t])
+
+  return {
+    videos: videos ?? [],
+    getVideo,
+    addVideo,
+    updateVideo,
+    setVideos,
+    setVideo,
+    removeVideo: removeProviderVideo
+  }
+}

src/renderer/src/hooks/video/useVideo.ts

@@ -0,0 +1,7 @@
+import { useProviderVideos } from './useProviderVideos'
+
+export const useVideo = (providerId: string, id: string) => {
+  const { videos } = useProviderVideos(providerId)
+  const video = videos.find((v) => v.id === id)
+  return video
+}

src/renderer/src/hooks/video/useVideoThumbnail.ts

@@ -0,0 +1,86 @@
+import { loggerService } from '@logger'
+import { retrieveVideoContent } from '@renderer/services/ApiService'
+import ImageStorage from '@renderer/services/ImageStorage'
+import { getProviderById } from '@renderer/services/ProviderService'
+import { Video } from '@renderer/types'
+import { useCallback } from 'react'
+
+const logger = loggerService.withContext('useRetrieveThumbnail')
+
+const pendingSet = new Set<string>()
+
+export const useVideoThumbnail = () => {
+  const getThumbnailKey = useCallback((id: string) => {
+    return `video-thumbnail-${id}`
+  }, [])
+
+  const retrieveThumbnail = useCallback(
+    async (video: Video): Promise<string> => {
+      const provider = getProviderById(video.providerId)
+      if (!provider) {
+        throw new Error(`Provider not found for id ${video.providerId}`)
+      }
+      const thumbnailKey = getThumbnailKey(video.id)
+      if (pendingSet.has(thumbnailKey)) {
+        throw new Error('Thumbnail retrieval already pending')
+      }
+
+      pendingSet.add(thumbnailKey)
+      try {
+        const cachedThumbnail = await ImageStorage.get(thumbnailKey)
+        if (cachedThumbnail) {
+          return cachedThumbnail
+        }
+
+        const result = await retrieveVideoContent({
+          type: 'openai',
+          provider,
+          videoId: video.id,
+          query: { variant: 'thumbnail' }
+        })
+
+        const { response } = result
+        if (!response.ok) {
+          throw new Error(`Unexpected thumbnail status: ${response.status}`)
+        }
+
+        const blob = await response.blob()
+        if (!blob || blob.size === 0) {
+          throw new Error('Thumbnail response body is empty')
+        }
+
+        const base64 = await new Promise<string>((resolve, reject) => {
+          const reader = new FileReader()
+          reader.onloadend = () => {
+            if (typeof reader.result === 'string') {
+              resolve(reader.result)
+            } else {
+              reject(new Error('Failed to convert thumbnail to base64'))
+            }
+          }
+          reader.onerror = () => reject(reader.error ?? new Error('Failed to read thumbnail blob'))
+          reader.readAsDataURL(blob)
+        })
+
+        await ImageStorage.set(thumbnailKey, base64)
+        return base64
+      } catch (e) {
+        logger.error(`Failed to get thumbnail for video ${video.id}`, e as Error)
+        throw e
+      } finally {
+        pendingSet.delete(thumbnailKey)
+      }
+    },
+    [getThumbnailKey]
+  )
+
+  const removeThumbnail = useCallback(
+    async (id: string) => {
+      const key = getThumbnailKey(id)
+      return ImageStorage.remove(key)
+    },
+    [getThumbnailKey]
+  )
+
+  return { getThumbnailKey, retrieveThumbnail, removeThumbnail }
+}

src/renderer/src/hooks/video/useVideos.ts

@@ -0,0 +1,48 @@
+import FileManager from '@renderer/services/FileManager'
+import { useAppDispatch, useAppSelector } from '@renderer/store'
+import { removeVideoAction } from '@renderer/store/video'
+import { objectValues } from '@renderer/types'
+import { useCallback } from 'react'
+
+import { useVideoThumbnail } from './useVideoThumbnail'
+
+export const useVideos = () => {
+  const videoMap = useAppSelector((state) => state.video.videoMap)
+  const dispatch = useAppDispatch()
+
+  const { removeThumbnail } = useVideoThumbnail()
+
+  const videos = objectValues(videoMap)
+    .flat()
+    .filter((v) => v !== undefined)
+
+  const getVideo = useCallback(
+    (videoId: string) => {
+      return videos.find((v) => v.id === videoId)
+    },
+    [videos]
+  )
+
+  const removeVideo = useCallback(
+    (videoId: string, providerId?: string) => {
+      const video = getVideo(videoId)
+      if (!video) {
+        return
+      }
+      if (!providerId) {
+        providerId = video.providerId
+      }
+      // should delete from redux state, and related thumbnail image, video file
+      if (video.thumbnail) {
+        removeThumbnail(videoId)
+      }
+      if (video.fileId) {
+        FileManager.deleteFile(video.fileId)
+      }
+      dispatch(removeVideoAction({ providerId, videoId }))
+    },
+    [dispatch, getVideo, removeThumbnail]
+  )
+
+  return { videos, getVideo, removeVideo }
+}

src/renderer/src/i18n/label.ts

@@ -146,7 +146,8 @@ const titleKeyMap = {
   notes: 'title.notes',
   paintings: 'title.paintings',
   settings: 'title.settings',
-  translate: 'title.translate'
+  translate: 'title.translate',
+  video: 'title.video'
 } as const
 
 export const getTitleLabel = (key: string): string => {
@@ -172,7 +173,8 @@ const sidebarIconKeyMap = {
   knowledge: 'knowledge.title',
   files: 'files.title',
   code_tools: 'code.title',
-  notes: 'notes.title'
+  notes: 'notes.title',
+  video: 'video.title'
 } as const
 
 export const getSidebarIconLabel = (key: string): string => {
@@ -329,7 +331,8 @@ const builtInMcpDescriptionKeyMap: Record<BuiltinMCPServerName, string> = {
   [BuiltinMCPServerNames.fetch]: 'settings.mcp.builtinServersDescriptions.fetch',
   [BuiltinMCPServerNames.filesystem]: 'settings.mcp.builtinServersDescriptions.filesystem',
   [BuiltinMCPServerNames.difyKnowledge]: 'settings.mcp.builtinServersDescriptions.dify_knowledge',
-  [BuiltinMCPServerNames.python]: 'settings.mcp.builtinServersDescriptions.python'
+  [BuiltinMCPServerNames.python]: 'settings.mcp.builtinServersDescriptions.python',
+  [BuiltinMCPServerNames.didiMCP]: 'settings.mcp.builtinServersDescriptions.didi_mcp'
 } as const
 
 export const getBuiltInMcpServerDescriptionLabel = (key: string): string => {
@@ -339,12 +342,14 @@ export const getBuiltInMcpServerDescriptionLabel = (key: string): string => {
 const builtinOcrProviderKeyMap = {
   system: 'ocr.builtin.system',
   tesseract: '',
-  paddleocr: ''
+  paddleocr: '',
+  ovocr: ''
 } as const satisfies Record<BuiltinOcrProviderId, string>
 
 export const getBuiltinOcrProviderLabel = (key: BuiltinOcrProviderId) => {
   if (key === 'tesseract') return 'Tesseract'
   else if (key == 'paddleocr') return 'PaddleOCR'
+  else if (key == 'ovocr') return 'Intel OV(NPU) OCR'
   else return getLabel(builtinOcrProviderKeyMap, key)
 }