Quality-aware
Simple requests go to small models; hard reasoning goes to stronger ones. You set the trade-off.
Model routing · Early access
The right model for every request
The router weighs quality, cost, latency and your policies, picks a model per request and falls back when a provider is busy.
The problem
Why this matters
Sending everything to the strongest model is slow and expensive. Sending everything to the cheapest one gives poor answers to hard questions.
Where it standsEarly access This is part of OpenUno Chat. We will say here when it ships.
How OpenUno helps
How it works
Policy rules
Pin regions, forbid lanes for some users, or require your own keys for sensitive work.
Fallbacks
When a provider rate-limits or fails, the request moves to the next lane instead of failing.
Related
More capability solutions
AI gateway
A single, standard API for every model you use, with keys, identity, budgets and limits applied in one place.
Read moreCost control
Every request is logged with its model, tokens and cost. Budgets stop spend before it happens, not after the invoice.
Read moreBring your own keys
Already have a contract with a model provider? Plug its key in and the router treats it like any other lane, billed to you.
Read moreAI that belongs to everyone.
Start free in the browser, or help shape the open-source core. Schools, partners and teams: we would love to hear from you.