OpenAI's expensive new model claims to solve problems like a doctoral candidate

All summer long, AI enthusiasts have been looking forward to the next big release from OpenAI, after our first teases about a next-gen project called Strawberry. The model would supposedly offer some big improvements over GPT-4, specifically when it comes to tasks involving programming, mathematics, and logical reasoning. Understandably, we’ve been excited to check out what it can do, and while we just heard the other day that Strawberry could land within the next two weeks, it’s gone ahead and made its debut already, arriving as OpenAI’s o1-preview.

While the new name makes clear that this is still an early look at what the model can handle, initial results already look really impressive, demonstrating an almost human-like ability to explain the “thought” process behind the model’s responses. OpenAI credits this performance on the further processing steps the o1-preview dedicates to generating its answers.

Taking its time doesn’t come without a cost, though, and as The Verge points out, o1-preview pricing runs 3-4 times as expensive as it did with the GPT-4o model. Seeing as it’s positioned towards scientists, researchers, and engineers, those fields may have access to the budgets to justify the expense, but the rest of us might want to make do with OpenAI o1-mini.

The scaled-down OpenAI o1-mini is still supposed to be quite good at coding and math, while also running a heck of a lot faster than the full-blown edition — and that means it costs 80% less, too. Some ChatGPT clients can start accessing both models now, with availability expanding through next week, and rate limits will put a cap on excessive use, helping to ensure as many people can try it as possible.

At some point in the future, OpenAI says that it plans to make o1-mini available to ChatGPT Free users, but offers no specific timeline.