Model and Reasoning Controls

The model picker now provides clearer information about each model, including its provider, deployment region, capabilities, and available reasoning levels. Agents can use an automatic selection mode, while organization administrators can choose which models are available and set an organization-wide default.

For compatible models, Interloom can preserve reasoning context between turns and display concise reasoning summaries alongside agent actions. This improves continuity on longer tasks without filling the conversation with duplicated thinking steps.