reasoning model
A model trained to work through a problem in hidden or summarised reasoning tokens before it writes its answer.
In more detail
Reasoning models do well on multi-step maths, code and planning, at the cost of extra latency and output tokens, which are billed even when hidden. They need less "think step by step" prompting and often ignore or reject sampling settings; you steer them with a reasoning effort or thinking budget instead.
Related terms
- reasoning effort — A request setting that tells a reasoning model how much thinking to do before answering, trading cost and latency for accuracy.
- chain of thought — Prompting a model to write out intermediate reasoning steps before its final answer.
- LLM token — The unit of text a language model reads and writes: a word, part of a word, a punctuation mark or a byte sequence from the tokenizer's vocabulary.
Taught in
- Tokens And The Context Window — AI Engineering: LLM Apps, RAG & Agents
- How The Next Token Is Chosen — AI Engineering: LLM Apps, RAG & Agents
- Choosing A Model — AI Engineering: LLM Apps, RAG & Agents
- Zero-Shot And Few-Shot — AI Engineering: LLM Apps, RAG & Agents
- Reasoning Techniques — AI Engineering: LLM Apps, RAG & Agents
References
- OpenAI: Reasoning models guide (unsupported sampling parameters)
- Anthropic: Building with extended thinking