From the archive
OpenAI releases o1-preview and o1-mini for reasoning tasks
The o1 preview models spend more time reasoning, with initial access through paid ChatGPT plans and selected API accounts.
OpenAI released o1-preview and o1-mini on September 12, 2024. The models were designed to spend more time reasoning before returning an answer, particularly for mathematics, science, and coding problems.
Two initial models
o1-preview introduced the new model series. The smaller o1-mini offered a lower-cost option focused on reasoning and coding rather than broad general knowledge.
The initial release reached ChatGPT Plus and Team accounts through the model picker. Application programming interface (API) access began with accounts qualifying for usage tier 5, subject to rate limits.
A preview with fewer chat features
The launch versions did not offer all the features available with GPT-4o. Browsing and file or image uploads were absent, and the initial API lacked function calling and streaming.
The new models therefore added an option for difficult reasoning tasks, while GPT-4o remained the more feature-complete choice for many ordinary chat and application workflows.