Choose a model
Pick the chat model and the models used for image, video, audio, and music.
Rolit uses two kinds of model choice: the chat model that reasons and replies, and generation models that make images, video, audio, and music.
Chat model
Use the model selector next to the message box.
| Option | Description |
|---|---|
| Auto | Picks the right model for the task. |
| Fast | Quick replies at lower cost. |
| Flash | Fast reasoning, lightweight. |
| Deep | Best reasoning, slower. |
In Agent mode the selector shows Agent instead. See Chat and runs.
Generation models
Open Generation model preferences in the composer to choose a model per media type: Image, Video, and Audio & Music. Rolit's agent uses your choice whenever it decides to generate that kind of media during the conversation, whichever studio tool is active.
If a run is already active, changing a model applies to that run immediately. Otherwise it applies from your next message. Clearing a choice returns that media type to the default.
The available models and their credit costs come from Rolit's model catalog, so the list can change. Each model's credit cost is shown with it. As examples, the catalog has included:
| Media | Example models |
|---|---|
| Image | FLUX-1-schnell, Imagen 3, Imagen 4 |
| Video | Seedance-1.5-Pro, p-video, p-video-2-pro, Veo 3.1 (lite, fast, and standard) |
| Audio | Chatterbox, Gemini 3.1 Flash TTS |
| Music | ACE-Step |
Set project defaults
To set default models for every generation in a project, use Project settings → Default Generation Models.