OpenAI, the creator of ChatGPT, continues to evolve its technology. Last year, in addition to releasing the GPT-4o model, it also released the o1 family of models. The o1 models stand out thanks to greater reasoning ability. As OpenAI explained, it is a series of AI models “designed to spend more time thinking before responding”. o1 models are therefore more effective for solving scientific, mathematical, or computer coding problems. In addition to the base version, o1, OpenAI also offers o1 mini, a more cost-effective model that excels in certain specific areas, as well as o1 pro, an improved version of o1.
But while the o1 family of models is still new, OpenAI is already preparing the release of the o3 models, which will be even more efficient. The development of o3 was announced in December. In a publication, Greg Brockman, president and co-founder of OpenAI, even spoke of a breakthrough. And the good news is that the company has made huge progress. In a recent post, Sam Altman states that OpenAI has already finalized the o3-mini model and is now focusing on the launch process. This launch should take place in the coming weeks. The other good news is that taking into account user feedback, OpenAI will offer o3-mini both as an API for developers, and on its ChatGPT application. The technology should therefore be available to a large number of users upon its launch.
ChatGPT goes into productivity mode
While improving its technology, OpenAI also makes ChatGPT more convenient for users. As we mentioned in a previous article, the company recently announced the “Tasks” functionality on ChatGPT. This allows you to use the chatbot to create reminders or schedule recurring tasks. In other words, ChatGPT no longer just responds to prompts, but can help users be more productive.
It makes sense for OpenAI to launch this feature, to be more competitive. Indeed, for its part, Google already offers a multitude of extensions for its Gemini artificial intelligence. These extensions make it possible to use generative AI to carry out actions on the firm’s other services, including the Workspace productivity suite.
- In 2024, in addition to launching GPT-4o, OpenAI also launched the o1 family of models which has superior reasoning ability
- And the new generation, called o3, is already in preparation
- Sam Altman announces in a publication that one of the models of this new family, called o3 mini, is already finalized
- Launch as an API and on ChatGPT is coming soon
