Leverage Default Adaptive Thinking for Complex Logical Reasoning Tasks
Trust the built-in reasoning to handle complex logic without manual toggles.
Stop manually enabling 'Extended Thinking' and instead provide complex tasks that allow the model's 'Adaptive Thinking' to trigger automatically.
The Scenario
You have a difficult logic puzzle or a complex business strategy problem that requires deep reasoning rather than a quick surface-level answer.
Before & after
Users previously had to manually enable 'Extended Thinking' mode or use complex 'Chain of Thought' prompts, which often took 10–15 minutes of prompt engineering.
Claude Sonnet 5's 'Adaptive Thinking' automatically scales its reasoning depth to the task's difficulty, providing a logical solution in 1–3 minutes without manual toggles.
The Prompt
Analyze this [COMPLEX_PRICING_MODEL/LOGIC_PUZZLE] and identify any logical inconsistencies or edge cases. show your reasoning process clearly.
Adaptive thinking is now enabled by default in Sonnet 5. The model automatically determines when it needs to 'think' more deeply about a problem without requiring the user to set a budget or manually trigger a reasoning mode.
Source
Claude Platform - Claude Platform Docs"Three behavior changes apply when migrating: adaptive thinking is now on by default; manual extended thinking... is removed."