Configure Model Reasoning

Aider supports two different reasoning approaches:

Reasoning Effort (for models like o3-mini):

aider --model o3-mini --reasoning-effort high

Or during a chat:

/reasoning-effort medium

Options: low, medium, or high

Thinking Tokens (for models like Claude 3.7 Sonnet):

aider --model sonnet --thinking-tokens 8k

Or during a chat:

/thinking-tokens 4k

These settings control how much effort the model puts into solving problems before responding, with higher values potentially yielding better results for complex tasks.