One key, many models
Call every published model with one credential and one base URL.
Developers
Build with one API
Point any OpenAI SDK at FullRouter, add a FullRouter key and keep your existing request shape. Routing, failover and billing stay behind the gateway.
OpenAI-compatible · Anthropic Messages · Responses API · SSE streaming
cURLPOST /v1/chat/completions
curl https://api.fullrouter.com/v1/chat/completions \
-H "Authorization: Bearer fr_your_key" \
-H "Content-Type: application/json" \
-d '{
"model": "anthropic/claude-sonnet-4.5",
"messages": [{ "role": "user", "content": "Hello" }]
}'Why the gateway
Same request shape, more control
FullRouter extensions live under routing; standard OpenAI fields are untouched.
Health-aware routing
Explainable candidates, retries before the first streamed byte and explicit fallbacks.
Transparent billing
Credit is reserved before the call, settled from provider usage and kept in an immutable ledger.
Documentation
Go deeper
Full documentation, API reference and operational notes live in the docs.