OpenAI o1 model now powers ChatGPT, coming to free users as well
The company also plans to bring the o1-mini to free users
2 min. read
Published on
Share this article
Improve this guide
Read our disclosure page to find out how can you help MSPoweruser sustain the editorial team Read more
Key notes
- OpenAI launched OpenAI o1, a reasoning model powering ChatGPT.
- The o1 is more accurate but slower and pricier than previous models.
- It’s available to Plus, Team, and soon Edu and Enterprise users.
OpenAI has launched its first-ever model for reasoning abilities, the OpenAI o1, and it’s coming to power the popular ChatGPT chatbot.
The latest model—alongside the smaller, cheaper OpenAI o1-mini— is the first in a new series of “reasoning” models designed to handle complex tasks, such as coding and multistep problem-solving. While more accurate and capable than previous models, o1 is slower and more expensive to use and still does have some issues with hallucination.
“Both o1-preview and o1-mini can be selected manually in the model picker, and at launch, weekly rate limits will be 30 messages for o1-preview and 50 for o1-mini. We are working to increase those rates and enable ChatGPT to automatically choose the right model for a given prompt,” says the Microsoft-backed company.
OpenAI o1 will power ChatGPT. Paid Plus and Team users are already able to try the model out, with Enterprise and Edu users starting next week. The company also plans to bring the o1-mini to free users, although the exact date has not been announced.
Compared to the previous GPT-4o model, o1, which scores 78.2% in the MMLU benchmark, outperforms the latter in 54 out of 57 of the benchmark’s subcategories despite an 82.0% overall score.
It also greatly outperforms GPT-4o in areas such as math, where it solved 74% of problems on AIME (compared to GPT-4o’s 12%), and even outperformed PhD-level experts on GPQA Diamond, a benchmark in chemistry, physics, and biology.
User forum
0 messages