Unverified50% confidenceFactExact time
Anthropic 在 Claude 推理模型中引入了可调节的推理强度机制（Effort Level），控制模型在生成答案前花费在思维链上的计算预算
1
Sources
50%
Confidence
Long-term
Relevance
7/22/2026
First Seen
Sources
Related Claims
UnverifiedReasoning Effort本质上是控制模型思维链深度的参数，档位越高模型允许的内部推理步骤越多，延迟和成本也成比例上升76% similarUnverifiedClaude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74% similarUnverifiedAnthropic在API层面提供thinking参数控制开关以及budget_tokens参数设定思考过程可消耗的最大Token数量73% 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.73% similarVerifiedClaude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73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/594058API
curl https://kongchang.com/api/v1/knowledge/claims/594058MCP
get_claim(id=594058)