Skip to main content

First Startup

Start the server​

./start.sh

Or manually:

source venv/bin/activate
python main.py

Verify that it works​

Open another terminal and run:

# Verify that the server responds
curl http://localhost:11435/api/version

# List loaded experts
curl http://localhost:11435/v1/models

You should see a JSON response with your experts listed.

Real-time logs​

Logs are saved in logs/app.log. To tail them in real-time:

tail -f logs/app.log

Successful startup log example​

[INFO] l3mcore API Server starting...
[INFO] GenericRouter initialized with embedding model: intfloat/multilingual-e5-small
[INFO] Loaded 3 experts from config/experts.json
[INFO] Expert 'programmer' → ollama @ http://127.0.0.1:11434
[INFO] Expert 'writer' → api (openai/gpt-4o)
[INFO] Expert 'analyst' → local (onnx)
[INFO] Server listening on http://0.0.0.0:11435

Make your first request​

curl http://localhost:11435/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {"role": "user", "content": "How do I make a loop in Python?"}
    ],
    "stream": false
  }'

Check the logs — you will see something like:

[Router] label='programmer' score=0.94 (embedding match)
[ExpertDispatcher] Routing to ollama @ http://127.0.0.1:11434 model=qwen2.5-coder:7b
It works!

If you see the routing log and receive a response, l3mcore is working correctly. You can now connect Open WebUI, Continue, or any OpenAI-compatible client.