Guide
Call the AI gateway
Route a request through one endpoint with caching, budgets, and tracing.
1
Create an API key
Issue a key for your project and give it a per-key budget so a runaway loop can't burn a month's spend.
2
Send a request to the gateway
Call the single gateway endpoint with your prompt and chosen model. Behind it, repeated prompts are served from cache and a degraded provider fails over automatically.
3
Inspect cost and latency
Every request is traced end to end — model, tokens, latency, cost — so 'why is this expensive?' is something you click on, not guess at. Swapping models is a config change, not a code change.
Ready to build?
Create a free account and follow along.