• feat(Markdown): disable indented code blocks
• fix(AssistantService): add default settings configuration to assistant initialization
• fix(i18n): fix model name export help text
• fix(GeminiAPIClient): update abortSignal option and ensure userLastMessage is pushed to messages
• fix: Chat does not work properly when configuring multiple API keys
• fix(ApiService): correct enableWebSearch conditional logic error
• fix(models): Unexpected inability to disable image generation feature
• fix: add node-stream-zip for zip file extraction in install-bun script
• fix: remove duplicated deepseek-v3 in volcengine
• fix: 修复数据目录迁移的bug
• fix: update source language handling and persist user selection in TranslatePage component
• refactor(middleware): Add error property to CompletionResult and handle errors when checking API

• made their first contribution in

Full Changelog:

发行代号：御 代号释义： 「御」，象征掌控与守护，寓意本次版本对系统稳定性、安全性与用户体验的全面驾驭与提升。

Dark	Light

• 增加配置文件校验，修复从古老版本升级上来的"No such file or directory (os error 2)"错误
• 修复扩展脚本转义错误
• 修复 macOS Intel X86 架构构建错误导致无法运行
• 修复 Linux 下界面边框白边问题
• 修复 托盘 无响应问题
• 修复 托盘 无法从轻量模式退出并恢复窗口
• 修复 快速切换订阅可能导致的卡死问题

• 新增 window-state 窗口状态管理和恢复

• 优化 托盘 统一响应
• 优化 静默启动+自启动轻量模式 运行方式
• 升级依赖

Windows (不支持win7)：

• 安装版： | （因为便携版问题较多不再提供便携版）

macOS 11+：

• DMG： |

Linux：

• DEB： | (Debian系) 使用 apt ./路径 安装
• RPM： | (Redhat系) 使用 dnf ./路径 安装

• ba8963c58380b09e9e61ffb200b414c71e3b8240 fix: fix GetProxyCountry
• 4f5c4d7a25df2c8fa00898feb4ccbea36aa1fd21 fix: sync workflow
• 6303c46e2e4f927f1ba3f64b8345f535926b201a op: GetProxyCountry add ua
• c3022949890cb3fc03a470ec853b68bb37e88426 op: update release.yml
• 376c1af712de2e4c315b1f59922260b0f72dcec4 op: 优化位置获取逻辑
• 8e8d79a6c51250845a8b8b69d1fa571a79a46a4f op: 优化最近日志滑动逻辑
• 3fa984920a0b45b7e8dbdb98e96bf5ed69427c18 update mihomo
• 93aadc5c0120287c612769d824785a0758e1f40d update sub-store to 2.19.64

It’s all about smoothing out the wrinkles and enhancing flexibility in this update. Here’s the rundown:

• LLM Node Enhancements: We've ironed out issues with missing parameters for structured outputs; now your LLM node is more robust and prepared for all it can handle thanks to .

• Gemini LLM Support: For those integrating with Gemini, you'll now find support for the Gemini 0.2.x plugin running smoothly in your agent apps, as addressed .

• Markdown Button Fix: Sending messages through the Markdown button should now work flawlessly, thanks to .

• Login Rate Limit Clearing: After a password reset, we’ve made sure the login rate limits are cleared, preventing any unnecessary hurdles, addressed .

• App Info Update: For better transparency, the author_name is now part of the app info, thanks to .

• Documentation Link Fixes: We’ve generalized the method for retrieving documentation links to respect localizations and squashed those error link paths, thanks to insights from .

• Document Indexing Boundaries: A fix ensures document indexing is correctly bound to a session, reducing those pesky unbound errors, tackled .

• Conversation Panel Optimization: We've fine-tuned the width adjustment logic for the conversation-panel modal, ensuring a smoother user interface experience, improved .

• CI Tests with Oceanbase: Integration tests now run the VDB tests of Oceanbase using Docker Compose within CI, streamlining our testing procedures thanks to .

• Dead Code Cleanup: We’ve removed some obsolete code utilizing Vulture, keeping our codebase leaner and cleaner, courtesy of .

• Localized Documentation Updates: Documentation link updates now support localization across various components, enhancing accessibility for our global community, thanks to .

Keep those pull requests coming as we aim to make your user experience as seamless and intuitive as possible! 🌟

1. Back up your customized docker-compose YAML file (optional)

   cd docker
   cp docker-compose.yaml docker-compose.yaml.$(date +%s).bak

2. Get the latest code from the main branch

   git checkout main
   git pull origin main

3. Stop the service. Please execute in the docker directory

   docker compose down

4. Back up data

   tar -cvf volumes-$(date +%s).tgz volumes

5. Upgrade services

   docker compose up -d

1. Stop the API server, Worker, and Web frontend Server.

2. Get the latest code from the release branch:

   git checkout 1.4.3

3. Update Python dependencies:

   cd api
   uv sync

4. Then, let's run the migration script:

   uv run flask db upgrade

5. Finally, run the API server, Worker, and Web frontend Server again.

• fix remote ip header CF-Connecting-IP
• fix(llm_node): missing parameters for structure outputs
• 🐛 Fix(Gemini LLM): Support Gemini 0.2.x plugin on agent app
• fix: markdown button can't send message
• refactor(sqlalchemy_workflow_execution_repository): Use the max funtion for getting next_sequence_number.
• feat: add S3_USE_AWS env var to explicitly distinguish AWS S3 usage in plugin-daemon
• fix(auth): Clear login rate limit after password reset
• chore: app info add author_name
• refactor: generalize method for getting doc link respecting locale and fix error link paths
• docs: conv and user_id
• document indexing not bound to a Session
• fix: correct description for edu coupon
• fix(web): optimize conversation-panel Modal width adjustment logic
• test: run vdb test of oceanbase with docker compose in CI tests
• fix: delete some dead code using vulture
• Feat/add rag dev deploy
• fix: update documentation links for various components to support localization
• chore: rename workflow blocks
• chore: bump version to 1.4.3

• made their first contribution in

Full Changelog:

• feat: Plugin server supports k8s deployment and configures the default download URL of the plugin(, ,)
• fix: change auto generate se namespace to mcp
• fix(ai-proxy): fix bedrock Sigv4 mismatch
• fix: fix const McpStreamableProtocol spell mistake
• feat(ai-proxy): Adjust the streaming response structure to keep it consistent with the openai
• add mcp-server doc
• opt: unify the end-of-line markers in the MCP session filter.
• fix(ai-proxy): fix gemini provider missing finishReason
• docs: fix broken link in mcp-servers README_zh.md
• feat: support for wanxiang image/video generation in ai-proxy & ai-statistics
• feat: Add AI-based bidding information tool MCP service
• fix: When configuring an MCP server for SSE forwarding, the controller may crash
• feat: add default route support for wanx image&video synthesis
• feat(ai-proxy): add support for OpenAI Fine-Tuning API
• add release-notes of 2.1.4
• feat: add mcp-router plugin
• update envoy and istio
• release 2.1.5-rc.1

• made their first contribution in
• made their first contribution in
• made their first contribution in

Full Changelog:

This is the v1.70.1 release of rclone.

Full details of the changes can be found in .

优化反射方法中的变量命名 优化框架获取 Toast 默认延迟时间 优化框架获取 Toast 默认延迟时间的方式 修改 Toast 跨页面显示的实现方式 开放 Toast 策略中的 Handler 对象给子类