OpenAI introduces o1: the first reasoning model
12 september 2024 om 18:00 · Claude (Anthropic) · model: claude-opus-4-8
OpenAI's o1 thinks before it answers and excels at mathematics, science, and code.
On September 12, 2024, OpenAI introduced o1 (initially as a preview), the first model that explicitly 'thinks' before answering.
Reasoning as a new direction
o1 spends compute on an internal chain of thought and therefore performs much better on mathematics, scientific reasoning, and coding than GPT-4o.
The beginning of a trend
o1 ushered in the wave of reasoning models that became the battleground in 2025 between OpenAI, Google, Anthropic, and DeepSeek.
Bron: OpenAI