Free LLM API — No API Key, No Sign-up
Call a free LLM with one curl command: no API key, no account and no wallet. OpenAI-compatible, rate limited per IP, with paid models one 402 quote away.
Try it
curl https://api.jarvisclaw.ai/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{"model": "auto/free", "messages": [{"role": "user", "content": "Hello"}]}'Send no Authorization header at all. The free tier admits only requests that carry no credential, and a placeholder key is rejected with 401.
Limits
Rate limited per IP address, measured in gateway calls per minute. The free models are open-weight models served from a shared pool, so latency varies.
When you need a paid model
Ask for any other model with the same request. The gateway answers HTTP 402 with the exact price quote in USDC; sign it with a Base or Solana wallet and repeat the request, or use an account API key.
OpenAI-compatible
The endpoint speaks the OpenAI Chat Completions format, so any OpenAI SDK works by changing the base URL.
from openai import OpenAI
client = OpenAI(base_url="https://api.jarvisclaw.ai/v1", api_key="unused")
# The SDK always sends a key header; use plain HTTP (above) for the keyless free tier.Who it is for
- Prototyping an agent or a chatbot before choosing a paid model.
- Tutorials, workshops and demos where nobody should have to create an account.
- Scripts and CI checks that need a model to answer, not a particular model.
FAQ
Is it really free?
Yes. Requests with model auto/free and no credential are not billed. They are rate limited per IP so the shared pool stays usable for everyone.
Which model answers?
auto/free routes to an open-weight model from the free pool. The response names the model that answered, so you can see which one you got.
Can I use it from the browser?
Yes. The endpoint allows cross-origin requests, so a static page can call it directly with fetch.