API Reference

Quick start

Anthropic-compatible clients use the root base URL. OpenAI-compatible clients use the /v1 base URL.

export ANTHROPIC_BASE_URL=https://nixo.codes
export ANTHROPIC_API_KEY=your-key-here

export OPENAI_BASE_URL=https://nixo.codes/v1
export OPENAI_API_KEY=your-key-here

Authentication

Send your key with every API request. Use the header that matches your client format.

curl https://nixo.codes/v1/messages \
  -H "x-api-key: YOUR_KEY" \
  -H "content-type: application/json" \
  -d '{"model":"anthropic/opus-5-5","max_tokens":1024,"messages":[{"role":"user","content":"Hello"}]}'

Create and manage keys in your dashboard. Keep keys on the server and out of browser code.

POST /v1/messages

Anthropic Messages-compatible endpoint for Claude, Fable, and supported open model groups.

{
  "model": "anthropic/opus-5-5",
  "max_tokens": 1024,
  "messages": [{ "role": "user", "content": "Explain recursion in one sentence." }],
  "stream": false
}

Set "stream": true to receive server-sent events (Claude Code always streams). Streaming is metered: the request is billed from the final usage in message_start and message_delta. stream must be a boolean. Use output_config.effort where the selected model supports it.

POST /v1/responses

OpenAI Responses-compatible endpoint for GPT and supported Fable models. This is the preferred endpoint for Codex CLI and Desktop.

{
  "model": "openai/gpt-6-sol",
  "instructions": "You are a helpful assistant.",
  "input": "Explain recursion in one sentence.",
  "stream": false,
  "store": false
}

• Streaming is supported and metered: with "stream": true the request is billed from the usage in response.completed.
• Response storage is disabled.
• A default instruction is supplied when instructions is omitted.

POST /v1/chat/completions

OpenAI Chat Completions-compatible endpoint. Claude requests are translated to Anthropic Messages; other model families route to their matching provider pool.

{
  "model": "anthropic/opus-5-5",
  "messages": [
    { "role": "system", "content": "You are a helpful assistant." },
    { "role": "user", "content": "Hello" }
  ],
  "max_tokens": 1024,
  "stream": false
}

With "stream": true Nixo always sets stream_options.include_usage, so the last chunk before [DONE] carries usage and the stream is billed from it.

See the model catalog for IDs and endpoint compatibility.

GET /v1/models

Returns the models available to your API key in an OpenAI-compatible list format.

curl https://nixo.codes/v1/models \
  -H "Authorization: Bearer YOUR_KEY"

For model capabilities and supported endpoints, visit Models.

Errors

Errors use a JSON object with an error field. Status and type depend on the failure.