OpenAI unveils new o3 and o3 Mini models; Here are the details

Open has been dropping things after things with the 12 day event Shipmas. And on the last day of the event, the company has introduced two new reasoning models o3 and o3 mini. The models have only been teased as of now and the launch dates have not been revealed.

OpenAI CEO Sam Altman suggested that OpenAI plans to launch the o3 mini by end of January 2025 and the full version of the model after a while. The main reason behind the launch is that the OpenAI boss expect that the new models will easily outclass the old ones and get new investors on board along with increasing the user base.

OpenAI o3 details

As of now, OpenAI is also accepting applications from the researchers to take part in the testing of these models before they are available for public. The application processor for the external researchers will end on January 10, 2025. Keeping thing aside, OpenAI recently launched the o1 model that was dubbed Project Strawberry in the beginning.

Now, with all the latest updates, there are chances that the company is ditching o2 and directly moving to o3. Reports also suggested that the OpenAI has done that to avoid any copyright scenario with a British company that goes by the name O2.

Further, reports have also suggested that the o3 model has performed far better than the previous model. Additionally, the upcoming model has also surpassed its predecessor in coding tests by 22.8 percent. The model has also passed with flying colors in one of the toughest maths contest like AIME 2024.

The model passed the exam with 87.7 per cent in the expert level science problems. Talking about the toughest math and reasoning problems, o3 was able to solve around 25.02 per cent of the same. As of now, the internal safety testing of the models is going on and they will be ready for market pretty soon.