Glossary
Model Hub
The gateway and router that picks a model for every AI call by measured evals, cost, latency and data policy.
The Model Hub sits between every AI feature and the models behind it. It is planned on an open-source proxy with a thin One3 router on top. Routing considers the task class, the data classification (restricted data goes only to local or in-region models), measured eval scores per task and model, cost and latency budgets, and a confidence cascade that tries the cheapest qualifying model first and escalates only when needed.
It provides per-tenant semantic caching (never across tenants), provider prompt caching, a batch queue, BYOK, fallbacks and per-call cost records. Swapping a model is a config change plus an eval run, never a code change.
Related terms
See how One3 would run your business
Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.