Verified80% confidenceFactExact time
Claude的Extended Thinking功能开启后，模型会在生成最终回答前先输出一段内部推理过程（thinking块），这段推理对用户可见但不计入对话上下文
11
Sources
80%
Confidence
Long-term
Relevance
5/28/2026
First Seen
Sources
用GPT自动配置Claude Opus 4全流程教程
bilibili如意-浏览器开发培训5/28/2026
Related Entities
Related Claims
UnverifiedClaude的扩展思考（Extended Thinking）模式会让模型在输出前进行显式内部推理链，提升复杂任务质量，但Token消耗量大幅增加82% similarVerifiedExtended Thinking is a reasoning enhancement mechanism introduced by Anthropic in the Claude 3 series that allows the model to perform an explicit internal Chain-of-Thought before generating a final answer.79% similarUnverified推理模型（如OpenAI的o1/o3系列、Anthropic的Claude with extended thinking、DeepSeek-R1）在生成最终答案前会先输出思考过程（thinking tokens）78% similarUnverifiedExtended Thinking的推理过程对用户可见，但不计入对话上下文77% similarUnverifiedClaude Code的Extended Thinking机制是ReAct框架的增强版本，允许模型在隐藏的思维链空间内进行更深层的多步推理77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1919API
curl https://kongchang.com/api/v1/knowledge/claims/1919MCP
get_claim(id=1919)