// live inference
Frontier models, served at the edge in under 40ms.
One endpoint. Twelve models. Automatic fallback, caching and routing — with zero infrastructure to babysit.
Tokens / sec
1,284,920
▲ 4.2% vs 60s ago
Requests · last 24h
1.24M
Peak 92K/sp50 38ms
Any model, one API
Swap Opus for Llama with a one-line change — routing and fallback handled for you.
OpusLlamaMixtral
Deploy your first agent in 60 seconds.
No credit card. Your first 1M tokens are on us.
Edge network
Live in 38 regions worldwide.
Technique: display:grid with named grid-template-areas + multi-track spans, re-mapped per breakpoint, plus a pointer-tracked radial glow