Real prices, real latencies, updated live — no sales deck, no "contact us". The catalog is the page. The page is the truth.
24 models · 6 labs · 0 markup · free tier included
| p50 latency | input / output | status | ||
|---|---|---|---|---|
| autorecommended | nexoai router | 380ms | best per request | live |
| claude-opus-5 | anthropic | 688ms | $5.00 / $25.00 | live |
| claude-sonnet-5 | anthropic | 412ms | $2.00 / $10.00 | live |
| gpt-5.6 | openai | 298ms | $1.25 / $10.00 | live |
| gemini-3-pro | 355ms | $1.25 / $5.00 | live | |
| deepseek-v3 | deepseek | 521ms | $0.27 / $1.10 | live |
| mistral-large-3 | mistral | 391ms | $2.00 / $6.00 | live |
Every number on this page is the number you'll be billed. Routing exists to make the table look better — latency down, cost down, uptime up.
Continuous probes from our edge to every provider. The ledger shows what the router knows — and the router always knows more than the table.
Lab list price, plus zero. The router's job is to spend your budget where the marginal token is cheapest — live.
The status column is a public API. Wire it to your own dashboard, page, or pager — trust the route, verify the route.
Build and prototype against every lab.
Lab prices, plus zero. The router spends it well.
Private routing, EU data residency, invoicing.
2 minutes to first request · usage-based · cancel anytime