Opencode2api

opencode2api

English · 简体中文

opencode2api is a local-first HTTP proxy that forwards OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages–style requests to the OpenCode upstream. It adds model aliases, reasoning/thinking compatibility, SOCKS5 proxying, token usage accounting, and a lightweight admin panel — so any OpenAI/Anthropic-compatible client can talk to OpenCode without changes.
This project is not affiliated with OpenAI, Anthropic, or OpenCode. Respect the upstream terms of service and only run it in environments you are authorized to use.

Features

- OpenAI-compatible endpoints: /v1/chat/completions, /v1/models
- OpenAI Responses compatible endpoint: /v1/responses
- Anthropic Messages compatible endpoint: /v1/messages
- Streaming SSE conversion with token usage accounting
- Model aliases, reasoning-effort mapping, and force-disable-thinking
- Multi-tier auth routing: public / auto / zen: / go: prefixes
- SOCKS5 support: direct, fixed proxy, or round-robin
- Web admin panel: edit config, view stats, reload upstream sessions
- GitHub Actions: multi-platform release binaries (Linux / macOS / Windows / FreeBSD)
- GitHub Actions: multi-arch Docker image published to GHCR (linux/amd64, linux/arm64)
- Single Go dependency (lumberjack); ships as one static binary

Project layout

Build locally:

One-click install

macOS, Linux, FreeBSD:

Windows PowerShell:

The installer downloads the latest release, verifies its SHA256 checksum, and installs the CLI into ~/.opencode2api/bin (or ~\.opencode2api\bin on Windows). launch claude / launch codex still require the corresponding local CLI. For version pinning, install directories, supported platforms, and manual install, see docs/INSTALL.md (安装 / installation).

Quick start

Health check:

List…