OpenCode OpenAI Proxy

OpenCode OpenAI Proxy is a small local bridge that lets OpenAI-compatible clients talk to OpenCode Go or OpenCode Zen.

It accepts OpenAI chat completion requests on POST /v1/chat/completions and forwards them to OpenCode Go or Zen.

What It Does

• Serves a local OpenAI-compatible API on 127.0.0.1:11435.
• Accepts POST /v1/chat/completions in OpenAI format.
• Supports streaming and non-streaming responses.
• Exposes /health for quick checks.
• Exposes /v1/models with direct OpenCode models and optional OpenAI route model names.
• Reads model inventory from models-go.json or models-zen.json.
• Reads OpenAI model routing from routes-openai-go.json or routes-openai-zen.json.
• Keeps the real OpenCode API key in OPENCODE_API_KEY, outside the repo.

Why

OpenCode Go and Zen use the OpenAI chat completions format. Some coding clients expect a local OpenAI-compatible endpoint. This proxy sits between them:

Cursor / VS Code / Any OpenAI client -> localhost:11435 -> OpenCode OpenAI Proxy -> OpenCode Go / Zen

How Models Work

Clients can send either direct OpenCode model IDs or real OpenAI model IDs from the route file:

• Cursor can send deepseek-v4-pro, qwen3.6-plus, gpt-5.5, or any model exposed by /v1/models.
• Cursor can also send routed OpenAI names such as gpt-5.2, gpt-4.1, or o3.
• VS Code extensions send whatever model ID they're configured for.
• Any OpenAI-compatible client works.

Direct OpenCode model IDs are forwarded as-is. OpenAI route model IDs are mapped through the selected routes-openai-*.json file.

In short:

• Clients talk OpenAI format to localhost.
• The proxy accepts direct Go/Zen model IDs and routed OpenAI model IDs.
• No format conversion needed — both sides speak OpenAI.

Requirements

• Node.js 18 or newer.
• An OpenCode API key in OPENCODE_API_KEY.
• Subscription tier in OPENCODE_TIER: go (default) or zen.

Quick Start

git clone https://github.com/bigdata2211it-web/opencode-openai-proxy.git
cd opencode-openai-proxy

cp .env.example .env
# Edit .env and set OPENCODE_API_KEY.

export OPENCODE_API_KEY=<your-opencode-key>
# Optional: switch from Go to Zen subscription (default: go)
export OPENCODE_TIER=zen
node index.js

The proxy starts on http://127.0.0.1:11435 by default.

To use another port:

node index.js 11436

Health Check

curl http://127.0.0.1:11435/health

Client Setup

Point OpenAI-compatible clients at the local proxy:

export OPENAI_BASE_URL=http://127.0.0.1:11435
export OPENAI_API_KEY=sk-dummy

Use one of the model IDs returned by /v1/models.

OPENAI_API_KEY is only a local client placeholder here. It is not the real OpenCode key and it is not routed or exchanged for a secret. Many OpenAI-compatible clients require this variable to be non-empty, so sk-dummy is enough.

The real secret stays only in the proxy process:

export OPENCODE_API_KEY=<your-real-opencode-key>

Request flow:

OpenAI-compatible client
  OPENAI_BASE_URL=http://127.0.0.1:11435
  OPENAI_API_KEY=sk-dummy
        |
        v
local proxy
  OPENCODE_API_KEY=<your-real-opencode-key>
        |
        v
OpenCode Go / Zen

This proxy can be used as a local OpenAI-compatible provider for clients that support a custom base URL, including coding CLIs, IDE extensions, and local tools such as Codex, OpenCode, Cursor, Roo Code, Cline, Continue, aider, and LiteLLM.

The client only needs to know:

• Base URL: http://127.0.0.1:11435
• API key: any non-empty local placeholder, for example sk-dummy
• Model: either a direct OpenCode model from models-go.json / models-zen.json, or an OpenAI route name from routes-openai-go.json / routes-openai-zen.json

Examples:

qwen3.6-plus    # direct Go/Zen model
deepseek-v4-pro # direct Go/Zen model
gpt-5-codex     # routed OpenAI model name
gpt-4o          # routed OpenAI model name
o3              # routed OpenAI model name

From the client's point of view, this proxy behaves like an OpenAI provider on localhost. The real OpenCode key stays in the proxy process as OPENCODE_API_KEY; clients do not need the real OpenCode key.

Model Config

Edit models-go.json for OPENCODE_TIER=go or models-zen.json for OPENCODE_TIER=zen. These files contain direct OpenCode model IDs:

{
  "models": ["deepseek-v4-pro", "qwen3.6-plus"]
}

Edit routes-openai-go.json or routes-openai-zen.json to map real OpenAI model IDs to the active tier's models:

{
  "gpt-5.2": "deepseek-v4-pro",
  "gpt-4.1": "qwen3.6-plus"
}

Route targets must exist in the matching models-*.json. The shipped route files cover every tier model once, without duplicate targets. Common context suffixes are accepted for both direct model IDs and routed OpenAI IDs, for example deepseek-v4-pro[200k] or gpt-5.2[200k].

The shipped configs are based on the current OpenCode /v1/models endpoints:

• models-go.json covers OpenCode Go models.
• models-zen.json covers OpenCode Zen models.
• routes-openai-go.json maps real OpenAI model IDs to Go models.
• routes-openai-zen.json maps real OpenAI model IDs to Zen models.

Endpoints

• HEAD / and HEAD /v1 - connection checks.
• GET /health - proxy status.
• GET /v1/models - available direct models and OpenAI route names.
• POST /v1/chat/completions - OpenAI-compatible chat endpoint.

Environment

Create a local .env from .env.example, or provide the variables another way:

OPENCODE_API_KEY=<your-opencode-key>
OPENCODE_TIER=go          # or: zen

Do not commit .env; it is ignored by git.

Subscription Tiers

Tier	OPENCODE_TIER	Endpoint	Pricing
Go	go (default)	https://opencode.ai/zen/go/v1/chat/completions	$5 first month, then $10/month (flat)
Zen	zen	https://opencode.ai/zen/v1/chat/completions	Pay-as-you-go, no limits

Both tiers use the same API key. Go and Zen model lists are kept separately in their JSON files, because OpenCode can change each tier independently.

To switch tiers, change OPENCODE_TIER and restart the proxy.

Updates and Contact

For free AI tools, news, and project updates, subscribe to the Telegram channel:

• https://t.me/gigaitools

For direct questions or feedback, message:

• https://t.me/xoskaz

Notes

This project is intentionally small: one Node.js entrypoint, tier-specific model config files, and no external runtime dependencies. Supports both OpenCode Go (flat subscription) and OpenCode Zen (pay-as-you-go) via OPENCODE_TIER.

License

No public license has been selected yet. The repository is public, but reuse rights are not granted automatically.