New in 2025–2026New 2024Builds on other terms
Reasoning model
Written by Minh Khang
What is Reasoning model?
A reasoning model is an AI model that takes time to "think" through several steps before answering, so it handles harder problems better.
OpenAI released its o1 models in September 2024, then Google, Anthropic and DeepSeek released their own reasoning models: they draft the steps before giving the answer. Results are more reliable for work that needs calculation or planning, but slower and cost more tokens.
Reasoning model in a shop
No need for it to quote delivery fees. To plan stock for Tet from three years of sales, a reasoning model does clearly better.
Sources
The dated facts on this page come from the original sources below.
Keep learning beyond the glossary
Read next
- BasicsLarge language model (LLM)A large language model (LLM) is the text brain behind ChatGPT, Gemini and Claude: it predicts the most likely next word, one word at a time.
- BasicsTokenA token is a small piece of text that AI reads and writes, such as a word, part of a word or a punctuation mark. AI prices and limits are counted in tokens.
- New in 2025–2026New 2025Context engineeringContext engineering is choosing the right information to give an AI at each step: enough to do the job, nothing extra to confuse it.