How do I adjust reasoning in the Playground?

Some Meta Model API models can reason through a problem before answering. When you select a reasoning-capable model in the Playground, a Reasoning effort control appears in the model settings panel.

Use it to choose how much effort the model spends reasoning (for example, low, medium, or high):

• Lower effort returns an answer faster and uses fewer tokens.
• Higher effort lets the model work through harder problems more thoroughly, which can use more tokens and take longer.

If you don't see a Reasoning effort control, the model you've selected isn't a reasoning model — switch models using the Model selector.

Higher reasoning effort can increase token usage, which affects what you're billed. For how reasoning works in the API, see Reasoning with the Responses API and Models in the developer documentation.