Releases: coaidev/coai
Releases · coaidev/coai
v4.0.0
CoAI v4.0.0 Release Notes
We are excited to announce the official release of CoAI v4.0.0! This is a major, milestone update focused on the refactoring of the backend management system and the enhancement of core data analysis capabilities, alongside numerous features and fixes contributed by the community.
🚀 Highlights
📊 Comprehensive Data Analysis Capabilities
This update introduces a brand-new backend data analysis module, elevating the statistical granularity from daily to minute-level. This provides finer, more real-time insights for operations and management.
- Minute-Level Real-Time Statistics: Added minute-level statistics for model request volume (RPM - Requests Per Minute) and token usage (TPM - Tokens Per Minute), enabling real-time monitoring for both individual users and the global system.
- Extended Billing and Request Statistics: New APIs for fetching yesterday's and last month's billing data, as well as statistics for error requests and total requests, ensuring comprehensive data coverage.
🏗 Admin Module Refactor
To support future feature expansion and improve code quality, we have thoroughly modularized and refactored the core Go backend admin/analysis module.
- Optimized Code Structure: The original
admin/analysis.gofile has been split into multiple files, includinganalysis.go,format.go,statistic.go, andreflect.go. This separation of concerns significantly enhances maintainability and readability.
✨ New Features & Improvements
- Skylark Integration: Merged Skylark-related features (PR #328), further expanding the platform's capabilities.
- Mathematical Formula Rendering: Added math preprocessing for the Markdown component, allowing for better display and rendering of complex mathematical expressions.
- Theme Switching Optimization: Optimized the display logic for theme switching, improving the user experience.
- Development Environment Configuration: Updated
docker-compose.watch.yamlto optimize the configuration for development and debugging environments.
🐞 Key Fixes
- Chat Content Display Fix: Fixed an issue where users could not see the answer content when the message included
reasoning_content. - Thinking Block Display Fix: Fixed the issue where text overflowed and was not displayed correctly in the "Thinking Block" due to the
overflow-hiddenproperty.
Detailed Changes
| Type | Module | Description | PR |
|---|---|---|---|
| Feat | Admin/Analysis | Added real-time statistics APIs like GetTpmToday and GetRpmToday. |
- |
| Feat | Admin/Analysis | Added GetBillingYesterday and GetBillingLastMonth APIs. |
- |
| Feat | Core | Merged Skylark-related features. | #330 |
| Feat | Frontend | Added math preprocessing for the Markdown component. | #335 |
| Refactor | Admin/Analysis | Split admin/analysis.go into analysis.go, format.go, statistic.go, reflect.go. |
- |
| Refactor | Frontend | Optimized the display logic for theme switching. | #350 |
| Fix | Chat | Fixed issue where answer content was hidden with reasoning_content. |
#326 |
| Fix | Chat | Fixed text overflow issue in the "Thinking Block" caused by overflow-hidden. |
#355 |
| Docs | Documentation | Added Japanese README documentation. |
#262 |
| Chore | Config | Updated the docker-compose.watch.yaml file. |
#324 |
| Chore | Config | Updated .gitignore to accommodate new config and key file structure. |
- |
| Chore | Dependencies | Major updates to package.json and pnpm-lock.yaml. |
- |
🤝 New Contributors
A big thank you to the following community members for their first contributions to CoAI! Welcome to the CoAI family!
Full Changelog: v3.11.1...v4.0.0
Thank you to all contributors and community members for your support! If you encounter any issues, please feel free to report them via GitHub Issues.
The CoAI Team
Release Date: 2025.10.23
3.11.1
What's Changed
- fix enter send message on macos by @yongman in #151
- claude sys msg adapt by @Lavanille777 in #152
- fix: skip the first claude system message by @Lavanille777 in #154
- 支持MySql tls连接 by @navee in #168
- feat: support model
gpt-4oandgpt-4o-2024-05-13by @XiaomaiTX in #196 - feat: add
gpt-4-turbogemini-1.5-pro-latestgemini-1.5-flash-latestmodel by @XiaomaiTX in #198 - fix: remove repetitive quota refresh interval by @eddsi in #180
- bug report: revert pr #180 "fix: remove repetitive quota refresh interval" by @zmh-program in #203
- chore: Define sending defaults based on different device types (#204) by @Sh1n3zZ in #211
- chore: Synchronizing updates from the
mainbranch by @Sh1n3zZ in #212 - feat: SMTP Compatibility Enhancement and SSL Protocol Support (#174) by @Sh1n3zZ in #213
- chore: Compatible with pre-modified SMTP sending method (#174) by @Sh1n3zZ in #214
- fix: fix used of network error on forced shutdown by @XiaomaiTX in #223
- feat: support the latest OpenAI model
chatgpt-4o-latestby @PeterDaveHello in #245 - Simplify apk upgrade process in the Dockerfile by @PeterDaveHello in #246
- Add a basic zh-tw i18n locale by @PeterDaveHello in #244
- feat: support spark desk new model by @mercury7720 in #253
- feature: add Alibaba Cloud Computing Nest one-click deployment button by @workhardliuzheng in #283
- fix: smtp starttls compatibility issue by @Ogannesson in #292
- Fix SMTP compatibility issues to support office365 provider by @zmh-program in #301
- fix: watchtower retries causing service crash by @zmh-program in #309
New Contributors
- @yongman made their first contribution in #151
- @Lavanille777 made their first contribution in #152
- @navee made their first contribution in #168
- @XiaomaiTX made their first contribution in #196
- @eddsi made their first contribution in #180
- @PeterDaveHello made their first contribution in #245
- @mercury7720 made their first contribution in #253
- @workhardliuzheng made their first contribution in #283
- @Ogannesson made their first contribution in #292
Full Changelog: v3.10...v3.11.1
Contributors
navee, PeterDaveHello, and 9 other contributors
Assets 2
3.10
v3.10
- feat: support unforced smtp registration 支持非强制 SMTP 发件注册 (#42)
- feat: support edit message and delete message 支持编辑和删除消息 (#51)
- feat: update file uploader and add auto copy file feature in chat input 支持输入框 Ctrl+V 复制图片 (#61)
- feat: support file viewer 支持上传文件查看 (#61)
- feat: suppport customize temperature, top_p, top_k params 支持自定义温度, TopP, TopK 等参数 (#61)
- feat: add is_banned field in user table 支持封禁和解封用户 (#61)
- feat: support update password, update email, set quota, set subscription level and release usage actions in admin user management interface 支持在用户管理中更新用户密码, 邮箱, 设置点数, 设置订阅等级和释放订阅用量 (#61)
- feat: support midjourney upscale / variation / reroll actions 支持 Midjourney U(放大)/V(变化)/R(重绘) 操作 (#40)
- feat: hiding secret feature 隐藏报错的模型密钥 (#67)
- feat: support unlimited plan usage 支持无限订阅配额 (设置配额为 -1)
- feat: support custom article/generation permission groups 支持自定义批量生成文章/项目生成器可使用分组
- feat: support model cache buffer feature 支持模型缓存
- feat: support close registration (#72) 支持关闭注册
- feat: support generate image and save image in sharing page 支持分享的对话的生成图片保存
- feat: support function calling in relay 中转支持 Function Calling
- feat: using tremor charts instead of react-chartjs 后台仪表盘图表优化: 使用 Tremor 替代 React ChartJs
- feat: optimize sharing feature, add mobile adapter, add maximized screen and minized screen feature 支持分享对话的全屏显示功能
- feat: support custom mask editor 支持自定义预设编辑 (#76)
- restruct: frontend conversation refactor and performance optimization 重构前端对话运行时和状态管理 (#82)
- feat: support rename conversation, support vision capability for claude-3 models and fix high context bug 修复高上下文文件仍会被自动切断的问题 (#83)
- fix: avoid truncating base64 images 修复 base64 图片会被识别为过长 自动切断问题
- feat: optimize scroll area and message acitons 优化对话滚动操作
- fix: fix subscription time pointer is nil error 修复订阅时间异常指针为空情况 (#90)
- fix: send key error 修复发送键问题 (#87)
- feat: readable subscription levels 可读化订阅等级 (#86)
- feat: seo optimization 自定义网站名称和 Logo 的 SEO 优化 (#91)
- feat: support markdown/html on site notify 通知支持 HTML/Markdown 格式
- fix: fix preset issues 修复预设在发送后消失问题
3.9
3.8
3.8
- ⭐ 新增模型市场（自定义设置模型名，模型请求 id，简介，标签，模型图像，参数，支持拖拽放置改变顺序，增删改查，自动绑定计费设置，自动同步更新）
- ⭐ 新增订阅管理（是否开启订阅，订阅价格，订阅物品 ID，显示名称，次数配额，勾选模型 - 自动绑定渠道已设置模型和常见模型列表，图标列表，上移下移，支持导入已设置的物品类型）
- ⭐ 新增服务日志（控制台，历史日志文件下载 查看，日志切片和定时清理 - 默认30d）
- 支持重置 API Key 功能
- 支持上传图片预览功能
- 优化移动端管理页面 UI
- 支持自定义文档地址
- 支持自定义文件解析服务
3.7
3.7
- ⭐ 支持 All in one 模式（当前可选 Deeptrain 模式 - 微信支付宝官方支付方式和 all in one 模式 - 发卡支付模式）
- ⭐ 支持渠道管理功能（优先级，权重，用户分组，模型映射，最大重试次数，接入点设置等）
- ⭐ 支持 DuckDuckGo API 快速联网搜索功能（未设置时使用较慢的 Bing / WebPilot 搜索逆向）
- ⭐ 支持自定义计费设置（支持按 Token 计费，按次计费，不计费，是否支持匿名调用等）
- ⭐ 支持模型市场模型拖拽放置，偏好模型顺序自定义
- ⭐ 新增后台系统设置（SMTP 发件，联网设置，自定义网站名称，网站 Logo 等）
- 更新偏好设置页面 （新增自定义发送键，最大历史会话数，语言设置）
- 新增日语语言适配
- 新增 Azure, Gemini 格式
- 增加鲁棒性响应截获
3.6
3.6
- 新增后台管理（仪表盘，用户设置，兑换码。公告设置等）
- 新增偏好设置
- 新增模型市场
- 新增预设功能
- 新增公告功能
- 新增订阅分层
- 新增批量文章生成功能
- 新增 GPT 4 Vision (gpt-4-vision-preview) 适配和图像 Token 计费
- 新增 Function Calling 适配
- 新增 DALLE-3 模型
- 新增 通义千问 模型
- 新增 ChatGLM3 Turbo 模型
- 新增 腾讯混元 模型
- 新增 360智脑 模型
- 新增 百川AI 模型
- 新增 抖音豆包 模型
- 新增 Stable Diffusion XL, LLaMa2, Code LLaMa 等 LocalLLM 模型
- 新增 Midjourney 模型（适配 Relax, Fast, Turbo模式）
- 更新 文件上传 功能（支持文本类型，pdf, pptx, docx, xlsx, 图片等格式）
- 更新命令行操作
- 优化兼容层重试/均衡负载机制
- 优化停止操作
- 优化文本输入和功能操作区
- 优化 Markdown 大文本渲染性能
- 修复订阅配额在请求报错时仍然计算问题
- 修复滚动反弹问题
- 修复 Safari 伸缩幅度过大问题
- 修复 Key 中转 Stream 流格式问题
- 删除 Service Worker 缓存机制（*因其在多数内核较老的浏览器内核中存在诸多问题，包括 PWA 自动更新等的兼容性问题较大）
- 支持桌面端 APP （基于 Tauri）
3.5
3.5
- 浏览器兼容性适配
- 更新模型选择器（标签，移动端适配）
- 分享对话管理功能
- 增加一键删除全部对话功能
- 修整冗余，前端广播触发策略更新
- 修复 Redux Dispatch 渲染上屏问题
- 修复 ChatGPT, Claude 卡字漏字问题
- 增加保留历史对话的使用模型功能
- 增加邀请码功能
- 增加侧边栏用户区域
- 更新缓存策略
- 更新 SSE 接收策略，提升并发速度
- 修复移动端提示通知无法自动消失问题
- 修复移动端重叠选择器重复触发问题
- 增加 Markdown XLink 功能
- 增加代码复制功能，以及右键区域复制功能
- 修复 Service Worker Manifest 被缓存而无法及时 PWA 更新的问题
- 增加订阅使用量情况信息显示
- 增加版本号显示和 PWA 自动安装到桌面功能（浏览器实验功能）
- 增加 Safari 浏览器支持
- 增加更便捷的移动端新建对话动作
- 更新 Claude 100k 用户配额策略
- 增加未发送输入框消息复原功能（防止误触刷新，pwa更新在某些浏览器上会强制刷新问题）
- 增加脚手架操作功能
- 增加 GPT-3.5 账号池自动筛选 Key 功能
3.4
3.4
- 对话分享功能
- 模型选择器更新
- 实现 停止，重新询问 等动作
- 增加对 chronium 66 ~ 84 的浏览器适配
3.3
3.3
- Key 中转服务
- Service Worker 自动更新功能
- Markdown 编辑器
- 新增模型适配：
- New Bing (unstable)
- Creative
- Balanced
- Precise