Fast answers aren’t always the best. That’s perhaps the most important lesson we’ve learned from the arrival of OpenAI Strawberry, now called o1-preview. It’s a new ChatGPT reasoning model that takes longer to give you potentially much better answers.
OpenAI announced the preview release on Thursday a blog poststating that it will appear in ChatGPT and the Generative AI company’s API. I can confirm that the o1-preview and a faster, cheaper model o1-mini are both live in our ChatGPT Plus account. The new models will not appear in the free ChatGPT accounts yet, however.
Strawberry has been hotly anticipated for its potential to think like humans. In the weeks leading up to this announcement, OpenAI CEO Sam Altman teased us with plenty of cheeky fruit references, but in recent months he’s also made it clear that generative AI would be taking a giant leap forward.
In the blog post, OpenAI explained: “We trained these models to spend more time thinking through problems before responding, just as a person would. Through training, they learn to refine their thinking, try different strategies, and recognize their mistakes.”
Ph.D intelligence
OpenAI claims that this more powerful o1-preview “performs comparable to PhD students on challenging benchmark tasks in physics, chemistry, and biology.” And that’s the key here. o1-preview is a generative model that may be most applicable in academia, not for helping you write a compelling prom-posal.
An example given in a video accompanying the blog is gene sequencing. In it, a scientist notes that humans can’t keep up with everything in gene sequencing, but an AI can. The scientist calls the new model “chat with reasoning” and shows how there’s a point where o1-preview says “Thinking” when she’s typing a question. Its value is that it keeps her from falling into the wrong part of gene theory.
O1-preview is not a replacement for ChatGPT-4o, which is barely a month old. The new model cannot search the web and cannot handle files and images. That will probably come at some point, though.
If biology and math aren’t your thing, the lighter and slightly more nimble o1 mini might be for you, and is now live in ChatGPT Plus. It’s particularly adept at coding.
You can try out the new models in ChatGPT Plus ($20 per month) by logging in and then selecting the model dropdown. You’ll notice that o1-preview and o1-mini have been added to the list at the time of publishing this story.
It will be interesting to see what people do with the model formerly known as Strawberry and whether people get impatient waiting for an o1 preview to provide answers. Maybe they will feel better if they can see that the extra reasoning results in more useful answers for some really big problems.