Glossary
chain of thought
A prompting and training technique where the model emits step-by-step intermediate reasoning before its final answer, improving accuracy on multi-step problems.
A technique with two lives. The original (Wei et al., 2022) was a prompting trick: append “Let’s think step by step” to elicit intermediate reasoning, improving accuracy on math and logic benchmarks. The follow-up (post-2024) is a training pattern: the “reasoning models” (o1, DeepSeekweightsA Chinese open-weight family known for the V3 MoE base model and the R1 reasoning model, both released under permissive licenses and unusually transparent in their training-cost reporting. Open full entry -R1, QwenweightsAlibaba's open-weight model family, leading the multilingual and Chinese-language open-weight space, released under Apache 2.0 with sizes from 0.6B to 235B parameters. Open full entry -R1, Gemini Thinking) are trained to emit long internal chains of thought that the user does not see, before answering.
The mechanism is partly externalizing intermediate computation: the model uses its own previous tokens as scratchpad. The mechanism is also partly capability transfer: training on chain-of-thought traces seems to instill general reasoning behavior usable on unseen problems.
In agent contexts CoT is usually combined with function callingagentsA pattern where a model emits a structured call (function name plus arguments), the runtime executes it, and the result returns as input on the model's next turn. Open full entry : the model reasons about which tool to call, calls it, reads the result, reasons about what to do next. ReActagentsAn agent loop where the model alternates between reasoning steps (thought) and acting steps (tool call), explicitly interleaving free-form deliberation with structured tool use. Open full entry (Yao 2022) formalized this loop; modern agent frameworks (GooseagentsBlock's open-source coding agent, BYOK across multiple model providers, with MCP support and a permissive license; the most cited fully-open agent platform in 2026. Open full entry , Claude Code’s harness, AutoGenagentsA Microsoft Research framework for multi-agent systems, with a conversation-pattern API for orchestrating multiple specialized agents to solve tasks collaboratively. Open full entry ) all use some variant.