Recommended model setup
Which models to enable in your workspace, which default model we recommend and how to set up both
Platform standard
As the default model for all members we recommend GPT-6.1 Terra or Claude Sonnet 5.5. DeepSeek-V4-Flash is an optional, particularly low-cost alternative for everyday use.
Models to enable
For all recommendations to work, these models should be enabled in the workspace:
| Provider | Models |
|---|---|
| GPT | GPT-6.1 Sol, GPT-6.1 Terra, GPT-6 Astra (DPA 2.0 required), GPT-6 Luna |
| Claude | Claude Sonnet 5.5, Claude Haiku 5.5, Claude Opus 5.5 |
| Gemini | Gemini 3.8 Flash, Gemini 3.5 Flash-Lite, Gemini 4 Argon |
| DeepSeek | DeepSeek-V4-Flash, DeepSeek-V4-Pro |
| Kimi | Kimi K3 |
Enable models
Open AI models
Open Settings → AI & features → AI Models.
Switch a model on
Each row is a model. Use the In workspace switch to turn it on or off.
Set who can use it
When you switch it on, the dialog asks who the model applies to: For all members or Only selected. The second option suits a pilot group, for example. More under Restrict models to specific teams or people.
Set the workspace default model
The default model applies to everyone who has not set their own.
Open the chat settings
Open Settings → AI & features → Chat.
Choose the default model
In the Standard model field, select the model you want.
Save
Click Save.
A personal default model that members set under Preferences takes precedence.
Inform your members
Share the recommendations under Models with your members. That page also explains how they get better results with Thinking before moving to a larger model. If a model is being retired, see Model lifecycle.