Choosing a model
Pick the AI model for a chat, compare models by their details, or let Auto choose.
Select the model name at the top of the chat to open Choose model. The list shows the models your organization offers you (see Organization Models): Auto first under Recommended, then the models grouped by provider. Type in Search models … to narrow the list.
Select a Language Model
- 1
Select the model name at the top of the chat. Choose model opens.
- 2
Select a model in the list to see its details. You can also move through the list with the arrow keys.
- 3
Select Apply or press Enter. If the chat already has messages, choose Use in this chat to continue the conversation with the new model, or New chat to start a new conversation with it.
New chats start with your default model, which you can choose in Personal Settings.
Model Details
For each model, the details show:
- Data protection: where the model runs (see the table below).
- Reasoning, Speed and Cost as a scale compared with the other models you can choose, and the Price per 1M tokens for input and output. If no price is set, AITEZA shows “No token price set”.
- Capabilities: whether the model Understands images, Searches the web and supports the Agent mode for datarooms and files.
- Whether it is the Default model of your organization.
| Label | Meaning |
|---|---|
| Self-hosted: maximum protection | The model runs on AITEZA’s own infrastructure or on dedicated servers of your organization. Your data never goes to an outside AI provider. |
| Hosted in the EU | The model runs at a provider in the EU. Your data is not used for training. |
| External API | The model runs through the provider’s API, possibly outside the EU. |
Models without a label have no hosting information yet.
Auto
With Auto, AITEZA picks a suitable model for every message, so you do not have to think about which model fits. Auto is offered as soon as you can choose at least two models.
How Auto decides
| Request | Examples | Model |
|---|---|---|
| Simple requests | Greetings, short questions, rephrasing | A fast, low-cost model |
| Everyday requests | Writing, summaries, explanations | Your organization’s default model |
| Demanding requests | Analyses, comparisons, code, long documents | The most thorough model |
The details of Auto in Choose model show which model is used for each kind of request.
Auto also takes into account what the message needs:
- Images go only to models that understand images.
- Web search goes only to models that can search the web.
- Connected datarooms limit the choice to the models the datarooms allow, and Auto prefers models with agent mode.
- Long documents go to models that can read a lot of text at once.
Within a conversation, Auto stays with the model it used before as long as that model fits, so answers keep a consistent style. It only switches to a stronger model when a message is clearly more demanding.
Below each answer, AITEZA shows Auto · model name. Point to it to see why Auto picked this model, for example “Auto picked Model X: demanding request”.
