Last week, OpenAI showed off its latest large language models (LLMs), OpenAI o1 and its little brother OpenAI o1-mini. In the announcement, the company said that Plus and Team subscribers could access the model from the announcement date. Enterprise and Edu subscribers would get it this week, and Free users would eventually get access to o1-mini. Now, it looks like Free users have access to o1-mini, with Neowin being able to confirm this as we have access.
To access the new model on the desktop, just press the ChatGPT Auto dropdown at the top left of ChatGPT then head to the Alpha Models option and you should see o1-mini if it has been enabled on your account. Once selected, the ChatGPT Auto dropdown with change to ChatGPT followed by the Greek letter for alpha to mark that you"re using an alpha model.
With regards to the mobile app, things are a bit different. If you long press on a ChatGPT response and press change model at the bottom of the context menu, there is now a new model listed called Alpha [gpt-4o]. It is not clear whether this is the o1-mini model or a more up to date version of GPT-4o.
For those who haven"t been keeping up, the new o1-mini and o1 models are able to spend more time thinking about what they want to say before they answer. This allows them to reason through complex tasks and solve harder problems related to science, coding, and math, compared to previous models.
According to OpenAI, these models perform similarly to a PhD student in physics, chemistry, and biology, based on benchmarks. It also did very well in a qualifying exam of the International Mathematics Olympiad (IMO), scoring 83% compared to GPT-4o"s 13%, and in Codeforces competitions it reached the 89th percentile.
Let us know in the comments if you have also been given access to o1-mini on a ChatGPT Free account.