arthu1/north-air-api

North Air 1 API

Optimized CPU inference with dynamic INT8 quantization.

Endpoints:

• —GET /health
• —POST /chat
• —POST /chat/stream

Request shape (/chat):

{
  "model": "north-air-1",
  "messages": [
    {"role": "user", "content": "Hello"}
  ]
}

Response shape:

{"output": "...", "inference": "pytorch-int8"}