Product
One platform. Every model. Total clarity.
Routing, caching, analytics and billing that work together from your very first request.
Deploy any model, serve any volume
One authenticated endpoint for every model in the catalog, with capability routing, fail-over, caching, rate limits and per-token pricing you see up front.
Explore →Every model you need, priced in plain sight
Browse every enabled model by capability, compare per-token prices before you call, and switch models without changing a line of integration code.
Explore →Prompt, run, see the price
Try any catalog model in the browser. See the response, token counts, latency and exact price for every run, then take the same request straight to production.
Explore →The right model, every request
Ask for a capability, get the lowest-priced enabled model that can serve it. If that model is busy, the request moves to another eligible model automatically.
Explore →Ask twice, pay once
Repeat an identical prompt inside the cache window and the answer comes back instantly from cache, with the same token counts and no charge for the call.
Explore →Every token, accounted for
Daily and month-to-date usage, per-request history with tokens, latency and price, and free allowance tracking. Know exactly where every token went.
Explore →Pay for tokens, not surprises
A free monthly token allowance, then per-token pricing shown per model before you call. Cache hits are free. Usage is invoiced every 30 days.
Explore →Controls you can see, keys we cannot read
API keys shown once and stored only as hashes, identity from verified credentials, per-account rate limits and a full request history for every call you make.
Explore →Inference for every team, governed from one place
One governed endpoint for every team: named keys, rate limits, request history, usage dashboards and clear per-token pricing. Talk to our team about scale.
Explore →