Send your first request

Send one test request after DeltaLLM is running and a model is available.

The commands use http://localhost:4002, the address for the Docker setup. If you used the development setup, replace it with http://localhost:8000.

This one local test uses the master key. Do not put the master key in an application.

1. Find your model name

curl http://localhost:4002/v1/models \
  -H "Authorization: Bearer YOUR_MASTER_KEY"

Replace YOUR_MASTER_KEY with the DELTALLM_MASTER_KEY value from .env.

The response lists the public names your requests can use. The Docker starter model is named gpt-4o-mini. If you chose a different public name, use that name in the next command.

2. Send a chat request

curl -X POST http://localhost:4002/v1/chat/completions \
  -H "Authorization: Bearer YOUR_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [
      {"role": "user", "content": "Hello!"}
    ]
  }'

Replace gpt-4o-mini if your model has a different public name.

A successful response contains the model's answer in choices[0].message.content.

Next step

Create an application key before connecting application code. An application key can be limited or revoked without affecting the rest of DeltaLLM.