Conceptual
Login

Chain-of-Thought Prompting

Eliciting intermediate reasoning steps from LLMs; why generating stepwise rationales improves multi-step task accuracy.

Why this Concept is marked deprecated

Superseded by Controlling Reasoning Effort Instead of Scripting the Reasoning: frontier models reason internally, and vendor guidance now says typed 'think step by step' is redundant and can degrade r

Read the source