Skip to content
Platform

Manage models

As an administrator, you configure the basic settings in the upper section of the model page and then activate or deactivate the available AI models.

Configure regions and default models

Allowed model regions

Select the hosting regions for AI models that are allowed for your account. You can select one or more regions. This setting does not apply to image and embedding models.

Available regions:

  • CH (Switzerland)
  • EU (Europe)
  • US (USA)
  • Other (other regions)

For example, if you want to use only Swiss models, select CH exclusively.

Fallback region

If your default AI model is unavailable, the system uses a fallback model from this region. This ensures your data is processed in your preferred region.

Available fallback regions: Switzerland, Europe, USA.

Default language model

This is the default language model that generates text and is preselected when creating a new agent. The model can be customized for each agent.

Default image model

This is the default model for image generation, used by agents when the plug-and-play image tool is activated. Custom image tools with their own models are not affected by this setting.

Default embedding region

This region defines which embedding model is used by default for new data sources. You can choose between Switzerland and USA.

Manage model cards

Below the configuration, all available AI models are displayed as cards – around 30 models, grouped by region: Switzerland, Europe, USA, and other regions.

Activate and deactivate models

Each model has a toggle in its model card. Click to activate or deactivate a model. Deactivated models are no longer available in your account and cannot be selected when configuring agents.

You can also turn all models of a region on or off together – for example, to deactivate all models outside Switzerland with a single click.

Show model details

Click Show details to see further information about a model:

  • Large Language Model (name)
  • Context window (size)
  • Model size
  • Credit costs per 10,000 tokens, split by input and output tokens

Credits and usage

The use of AI models is based on credits. Credit consumption depends on the model used and the number of tokens processed. Costs are transparently displayed per model and split by input and output tokens.