Skip to main content
You can choose model variants that change reasoning effort, processing speed, or both. Use reasoning effort to adjust how much work the model puts into a response. Use a fast service tier when you want faster processing and accept higher pricing. The available controls depend on the model family and your workspace restrictions.

How to choose a model and a reasoning effort

  1. Open the model picker
  2. Find a model family
  3. Open Configure variants
  4. Choose an effort
  5. Check the model
  6. Send your message

​
Choose a variant

  1. Open the model picker beside the message composer.
  2. Find a model family and open Configure variants beside its name.
  3. If Mode appears, choose from the offered modes: Standard, Thinking, or Pro. Thinking uses extended reasoning. Pro uses additional model work for difficult tasks.
  4. If Speed appears, choose Standard or Fast. Fast uses faster processing at higher pricing. The picker stays open so you can finish choosing effort.
  5. Under Effort, choose an available option. Choosing an effort or a mode closes the picker. Check the selected variant in the composer before sending your next request.

​
Understand effort choices

You may see the following labels. Each family offers its own subset.

​
Good to know

Effort and speed are separate choices. Low changes reasoning effort; Fast changes processing speed and pricing. You can combine them only when the family offers that combination. You see Mode or Speed only when there is more than one available choice. A missing control does not mean you need another permission. The model may not support it, or your workspace or deployment may exclude its variants. Fast processing is unavailable on endpoints that cannot provide that service tier.