Skip to content

Server and Client

The entire Lumos API is also available as a server / client fashion to deploy and remotely call the API.

A seperate service can be useful isolation for

  1. Running compute heavy book parsing operations
  2. A centralised AI server for your org

Deploy (Server)

Simply host the FastAPI server, authenticated by an API key: 

LUMOS_API_KEY=12345678 
uv run uvicorn lumos.server.app:app --host 0.0.0.0 --port 8000

Client SDK

Once deployed, you can conveniently access the service with the LumosClient that fully mirrors the Python API. 

from lumos import lumos
can be safely replaced with 
from lumos import LumosClient
lumos = LumosClient(host="http://localhost:8000", api_key="12345678")

Now you can do similar operations like: 

from pydantic import BaseModel

class Response(BaseModel):
    steps: list[str]
    final_answer: str


lumos.call_ai(
    messages=[
        {"role": "system", "content": "You are a mathematician."},
        {"role": "user", "content": "What is 100 * 100?"},
    ],
    response_format=Response,
    model="gpt-4o-mini",
)
# Response(steps=['Multiply 100 by 100.', '100 * 100 = 10000.'], final_answer='10000')