Reasoning model
GPT-4.5.icon
The Reasoning model, a large-scale language model (LLM), is a type of LLM designed to effectively handle complex inference tasks. While traditional LLMs generate responses by predicting the next word based on statistical patterns, the Reasoning model decomposes a problem into multiple steps and performs step-by-step inference to produce a more accurate answer.
For example, OpenAI's "o1" model released in September 2024 and the "o3-mini-high" model released in December 2024 have shown better performance than traditional LLMs in math, science, and programming. In addition, in January 2025, DeepSeek, a Chinese company, released "DeepSeek-R1", a 671 billion-parameter inference model that achieves performance comparable to OpenAI's "o1" model. These Reasoning models require more computational resources for each query than traditional LLMs, but they perform better on tasks that require mathematical reasoning and logical thinking.
---
This page is auto-translated from /nishio/Reasoningモデル using DeepL. If you looks something interesting but the auto-translated English is not good enough to understand it, feel free to let me know at @nishio_en. I'm very happy to spread my thought to non-Japanese readers.