Extended thinking
Step-by-step reasoning the shows before it answers. See what it is, how it works, how it is used and what to watch for, with pictures and examples.
What it is
Extended thinking is the step-by-step reasoning writes before it gives its answer. It helps Claude work through a problem before acting. On with adaptive reasoning, the effort level is the main control for how much thinking happens, while the thinking settings turn it on or off and change how it is shown. folds the thinking away by default so your screen stays tidy.
How it works
-
You send a
-
Claude reasons step by step
-
Claude Code folds the thinking away
-
Claude writes the final answer
How to use it
-
Option+T() orAlt+T(, )turns thinking on or off for this session.
-
/configsets the default, saved as
alwaysThinkingEnabledin~/.claude/settings.json. -
Ctrl+Otoggles verbose mode so you can read the thinking as gray italic text.
-
showThinkingSummariesset it to
truein settings to get full summaries when you expand.
Watch out
You pay for every thinking , even when the thinking is folded or redacted. Thinking cannot be turned off on Opus 5.5, Sonnet 5.5, Haiku 5.5 or Fable models; there the model decides per step based on the effort level. MAX_THINKING_TOKENS caps thinking only on models with a fixed .
Related terms
Original and official docs
This page explains the term in plain words. The original is the authority: if they differ, follow the original.
- See this term in the originalClaude Code Docs — Glossary Anthropic · checked 2026-10-19
- Anthropic — Claude Code docs: Model configuration — Extended thinking(opens in a new window)
- Anthropic — Claude Code docs: Model configuration — Adaptive reasoning and fixed thinking budgets(opens in a new window)