What is Adaptive Reasoning?

Adaptive reasoning is a new feature in OpenAI’s GPT-5.1 models (released November 2025) that allows the model to automatically adjust the “depth” of its thinking process based on the complexity of the submitted prompt. This feature helps the models balance speed and thoroughness, providing quicker responses for simple questions and engaging in deeper, more complex reasoning when necessary.

Where can I find it?

Adaptive reasoning is available by default in both the GPT-5.1 Instant and the GPT-5.1 Codex and Codex Mini models. All models are tagged accordingly.

Didn’t find what you needed? Please reach out to research.computing@dartmouth.edu.

How do I set the degree of "reasoning" when using REASONING models?

How do I use the Dartmouth Chat API?

On this page: