Five-minute quickstart
Keep your SDK. Change the endpoint and key.
Connect a provider credential, mint a gateway key, and send the same OpenAI-compatible request shape your application already uses.
- 1. Connect a provider
Add your own provider credential. It is validated before it can serve traffic.
- 2. Create a gateway key
The full key is shown once. Only its hash is retained.
- 3. Send a request
Use the gateway URL as your SDK base URL and prefix models with the provider.
OpenAI-compatible request
One endpoint for governed traffic
The gateway key authenticates your application. The connected provider credential is injected only for the selected upstream request.
curl https://api.tensorproxy.com/v1/chat/completions \
-H "Authorization: Bearer gw_live_..." \
-H "Content-Type: application/json" \
-d '{
"model": "openai/gpt-4o-mini",
"messages": [{"role": "user", "content": "Say hello"}]
}'Model discovery
Call GET /v1/models with the gateway key to receive exactly the provider-prefixed models that key may use. The result is generated by the gateway and does not consume provider quota.