12 Days of OpenAI ends with a new model for the new year
- OpenAI has announced upcoming o3 and o3-mini AI models.
- The new models are improved “reasoning” AI models that build on the o1 and o1 mini models released this year.
- Both models handily outperform existing AI models and will be rolled out in the coming months.
On the final day of the 12 Days of OpenAI, OpenAI CEO Sam Altman returned to show off a brand new set of AI models coming in the new year. The o3 and o3 mini models are improved versions of the relatively new o1 and o1 mini models. They are designed to think before they speak and reason their answers. The mini version is smaller and more focused on performing a limited number of specific tasks, but with the same approach.
OpenAI calls it a big step toward artificial general intelligence (AGI), which is a pretty bold claim for what is in some ways a mild improvement on an already powerful model. You may have noticed that there is a number missing between the current o1 and the upcoming o3 model. According to Altman, this is because OpenAI wants to avoid any confusion with the British telecom company O2.
What makes o3 special? Unlike regular AI models that spit out answers quickly, o3 takes time to reason things out. This “private thinking” allows the model to fact-check itself before responding, helping it avoid some of the classic AI pitfalls, such as confidently uttering wrong answers. This extra thinking time may slow down o3, even if just a little, but the reward is better accuracy, especially in areas like math, science, and coding.
A nice thing about the new models is that you can manually adjust that extra thinking time. If you’re in a hurry, you can set it to ‘low compute’ for quick responses. But if you want top-notch reasoning, set it to ‘high compute’ and give it some more time to think about it. In tests, o3 has easily surpassed its predecessor.
This isn’t quite AGI; o3 cannot take over from people in all ways. It also doesn’t meet AGI’s OpenAI definition, which describes models that outperform humans on the most economically valuable projects. But if OpenAI achieves that goal, it will be interesting for the collaboration with Microsoft, because it would end OpenAI’s obligation to give Microsoft exclusive access to the most advanced AI models.
New year, new models
At this time, o3 and its mini counterpart are not available to everyone. OpenAI is giving security researchers a sneak peek through Copilot Labs, and the rest of us can expect the o3 mini model to release in late January, with the full o3 following shortly after. It’s a careful, measured rollout, which makes sense given the power and complexity we’re talking about here.
Still, o3 gives us a glimpse of where things are going: AI that not only generates content, but actually thinks about problems. Whether it brings us to AGI or not, it’s clear that smarter, reasoning-driven AI is the next step. For now, we’ll have to wait and see if o3 lives up to the hype or if this latest gift from OpenAI is just a lump of coal in disguise.