API docs

Use the API at https://api.chigyu.ai.

Endpoints

GET /health — gateway health

Gateway liveness. Should return "ok", otherwise the server is down.

curl https://api.chigyu.ai/health

Response · 200

{"status": "ok"}

GET /endpoints — LLM endpoint availability

LLM endpoint availability. Serving engines are swapped out frequently and sometimes there might be multiple online, this will list them.

curl https://api.chigyu.ai/endpoints \
  -H "Authorization: Bearer $CHIGYU_API_KEY"

Response · 200

{
  "endpoints": [{
    "name": "radiance-1",
    "model": "qwen3.8-27b",
    "online": true,
    "checked_at": 1791169353
  }]
}

GET /v1/models — online models

Standard OpenAI-compatible /models endpoint.

curl https://api.chigyu.ai/v1/models \
  -H "Authorization: Bearer $CHIGYU_API_KEY"

Response · 200

{
  "object": "list",
  "data": [{
    "id": "qwen3.8-27b",
    "object": "model",
    "created": 0,
    "owned_by": "lutetai"
  }]
}

Use the returned model ID in inference requests.

POST /v1/chat/completions — model serving

Standard OpenAI-compatible /chat/completions endpoint.

curl https://api.chigyu.ai/v1/chat/completions \
  -H "Authorization: Bearer $CHIGYU_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "qwen3.8-27b",
    "messages": [{"role": "user", "content": "Say hello."}],
    "reasoning_effort": "low",
    "max_tokens": 1024
  }'

Response · 200 (abridged)

{
  "object": "chat.completion",
  "model": "qwen3.8-27b",
  "choices": [{
    "index": 0,
    "message": {"role": "assistant", "content": "Hello!"},
    "finish_reason": "stop"
  }],
  "usage": {"prompt_tokens": 19, "completion_tokens": 2, "total_tokens": 21}
}

Streaming

curl -N https://api.chigyu.ai/v1/chat/completions \
  -H "Authorization: Bearer $CHIGYU_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "qwen3.8-27b",
    "messages": [{"role": "user", "content": "Say hello."}],
    "reasoning_effort": "low",
    "max_tokens": 1024,
    "stream": true
  }'

Response · 200 · text/event-stream (abridged)

data: {"choices":[{"index":0,"delta":{"content":"Hello!"}}]}

data: {"choices":[],"usage":{"prompt_tokens":19,"completion_tokens":2,"total_tokens":21}}

data: [DONE]

POST /v1/completions — text prompts

Standard OpenAI-compatible /completions endpoint.

curl https://api.chigyu.ai/v1/completions \
  -H "Authorization: Bearer $CHIGYU_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"model":"qwen3.8-27b","prompt":"The capital of France is","max_tokens":32}'

Response · 200 (abridged)

{
  "object": "text_completion",
  "model": "qwen3.8-27b",
  "choices": [{"index": 0, "text": " Paris.", "finish_reason": "stop"}]
}

Raw text completion; use chat/completions for conversations. Streaming is also supported.

Python · OpenAI SDK

pip install openai

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.chigyu.ai/v1",
    api_key=os.environ["CHIGYU_API_KEY"],
)

stream = client.chat.completions.create(
    model="qwen3.8-27b",
    messages=[{"role": "user", "content": "Say hello."}],
    reasoning_effort="low",
    max_tokens=1024,
    stream=True,
)

for chunk in stream:
    if chunk.choices:
        print(chunk.choices[0].delta.content or "", end="", flush=True)
print()

OpenCode V2

Add this provider to ~/.config/opencode/opencode.json.

{
  "$schema": "https://opencode.ai/config.json",
  "providers": {
    "chigyu": {
      "name": "chigyu.ai (lutetai)",
      "env": ["CHIGYU_API_KEY"],
      "package": "@opencode/ai/providers/openai-compatible",
      "settings": {"baseURL": "https://api.chigyu.ai/v1"},
      "models": {
        "qwen3.8-27b": {
          "name": "Qwen3.8 27B (chigyu.ai)",
          "settings": {"reasoningEffort": "xhigh"},
          "variants": [
            {"id": "low", "settings": {"reasoningEffort": "low"}},
            {"id": "medium", "settings": {"reasoningEffort": "medium"}},
            {"id": "high", "settings": {"reasoningEffort": "xhigh"}},
            {"id": "xhigh", "settings": {"reasoningEffort": "xhigh"}}
          ],
          "capabilities": {
            "tools": true,
            "input": ["text", "image"],
            "output": ["text"]
          },
          "limit": {"context": 262144, "output": 32768}
        }
      }
    }
  }
}

Use /connect to enter your API key, or supply CHIGYU_API_KEY to the OpenCode server.

Then select chigyu/qwen3.8-27b in /models.

The high preset maps to xhigh because this backend does not accept high.

Serving engines

We use various open source inference engines for model serving.

Currently, we are just using a slightly modified vLLM Radiance.