# Models and providers

> How Geenius Chat talks to AI providers through one interface, how a router picks the provider, where keys are kept, and what is live today.

Source: https://chat.geenius.io/docs/models/

Geenius Chat is not tied to one AI company. It talks to AI **providers** through one interface.

## One interface

Each provider is a plug-in behind the same interface. A **router** decides which provider answers a message. Because every provider looks the same to the rest of the engine, a new model can be added without changing how you chat, and without changing your threads or runs.

## Keys stay on the server

A provider needs a key. In Chat, keys are handed to the provider plug-in on the server when it needs one. They are never put in your browser or in the app's code. Bringing your own key is not offered.

## How we test providers

Every provider plug-in is tested against recorded examples. Our tests never call a real model. That keeps the tests repeatable and keeps keys out of them.

> [!HONEST]
> No real model call happens in Chat yet. The provider interface and the router are built and tested; the models you can pick are listed when conversations go live.

## Which model answers

When Chat opens, the app shows which models you can pick. It will say which provider answers, so you know where your message goes.

## Related

- [Your data](/docs/your-data/): what is sent to a provider.
- [Glossary](/docs/glossary/): provider, router, model.