auto

How auto picks a model

Each request goes to the model that has been answering best. A model that failed in the last 10 minutes waits its turn, and models under 80% uptime over the last 3 days go to the back.

If the model picked fails, the next one is tried, up to 3 models for one request. Only the reply you receive is charged, and a request that no model could answer does not count toward your calls.

Every reply names the model that wrote it.

Call it through the API

Any OpenAI-compatible client works: use the address in the sample and this model name.

import requests

response = requests.post(
    "https://onomeo.com/v1/chat/completions",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={"model": "auto", "messages": [
        {"role": "user", "content": "Hello!"}
    ]}
)
print(response.json())