v0.2.7
opencode-ai/opencodev0.2.7Sep 19, 2026by github-actions[bot]
Upgrading?
AI Summary
Adds native support for Volcano Ark Seedance video tasks and enhances the local plugin system with host services and status channels.
Key Highlights
- Native Seedance (Ark) video API with async task protocol and usage-based billing
- Plugin host service providing KV storage and account directory capabilities
- Read-only status channel for real-time plugin monitoring without re-authentication
- Optimization of mobile UI to retain model plaza icon entry
New Features
- Seedance video API support
- Plugin host service
- Read-only status channel
- Mobile UI optimization
Full Release Notes
> AI API Gateway Platform - 将 AI 订阅配额分发和管理 新增火山方舟 Seedance 原生视频任务 API,并为本地插件系统引入通用宿主服务与只读状态通道。 ## 新增功能 - Seedance (Ark) 原生视频 API:支持火山方舟异步视频任务协议(创建/查询/删除),按上游 usage 计费,响应保持原生格式 - 插件宿主服务:为本地插件提供按插件命名空间的 KV 存储、账号目录(受插件已声明能力范围限制)等通用能力 - 插件只读状态通道:新增 /admin/plugins/:id/status 与 plugin.status UI 通道,可无需二次验证轮询插件实时状态 ## 优化改进 - 手机竖屏顶栏保留模型广场图标入口 ## Bug 修复 - 修复 Antigravity Gemini 裸模型名(配合 thinkingConfig)被直发上游导致 404 的问题 - 修复 Antigravity Gemini 原生流对 go-genai / python-genai 客户端发送 SSE 注释心跳导致流中断的问题 - 修复 Kimi 国内 Coding Plan 账号配额耗尽(403)被误判为永久禁用的问题,改为按窗口重置时间限时暂停 - 修复 Codex 内置工具根级联合 schema(oneOf/anyOf/allOf)导致 /v1/responses → /v1/messages 返回 400 的问题 - 修复 DeepSeek 思考模式在 Chat 回退时缺少 reasoning_content 导致 400 的问题 --- ## 📥 Installation **Docker:** ```bash # Docker Hub docker pull weishaw/sub2api:0.2.7 # GitHub Container Registry docker pull ghcr.io/wei-shaw/sub2api:0.2.7 ``` **One-line install (Linux):** ```bash curl -sSL https://raw.githubusercontent.com/Wei-Shaw/sub2api/main/deploy/install.sh | sudo bash ``` **Manual download:** Download the appropriate archive for your platform from the assets below. ## 📚 Documentation - [GitHub Repository](https://github.com/Wei-Shaw/sub2api) - [Installation Guide](https://github.com/Wei-Shaw/sub2api/blob/main/deploy/README.md)
Related Repositories
ollama/ollama★ 184,209
Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
fatedier/frp★ 110,051
A fast reverse proxy to help you expose a local server behind a NAT or firewall to the internet.
infiniflow/ragflow★ 92,612
RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context layer for LLMs
gohugoio/hugo★ 90,198
The world’s fastest framework for building websites.
junegunn/fzf★ 83,716
:cherry_blossom: A command-line fuzzy finder