For developers
Your users bring their own AI.
Add a Connect with OpenLLM button. Your users' API keys, Kimi, Grok and local models power your app, with fallback built in. Inference stays off your bill.
Your app
Put the Connect with OpenLLM button where AI turns on.
OpenLLM asks
Your user approves once, on OpenLLM, within limits they set.
Their model answers
Requests run on the keys and models they already have, not on your bill.
Your bill doesn't grow with your users.
$0 inference
Every request runs on your user's own API key, local model or subscription. More users never means a bigger AI bill.
Every provider
API keys, local models and subscriptions, behind one OpenAI- and Anthropic-compatible endpoint.
Users stay in control
They set limits and expiry, and disconnect in one click. You never see their provider keys.
How it works
Every app is reviewed by a person before it goes live. It takes a short call, not a long form.
- 01
Apply
Tell us about your app: what it does, where it lives.
- 02
A short call
We review every app by hand and email you to set one up.
- 03
Approval
We approve the app and email you when it's ready.
- 04
Get your credentials
Settings → Developers → Apps → Set up. Made in your browser, kept in your vault.
- 05
Paste one prompt
Give it to your coding agent. It adds the button and the calls.
Just need an API key? Users paste their OpenLLM API key and you point your base URL at https://www.openllm.sh/v1.