The technology company OpenAI has just released an update for ChatGPT, introducing two new language models called o3 and o4-mini, which are expected to bring a different artificial intelligence (AI) experience compared to before.
Free users can also access these changes.
The update makes ChatGPT smarter and more responsive, especially in tasks requiring complex reasoning, compared to previous versions.
The first notable feature is ChatGPT's ability to intelligently decide which tool to use to address a user's request.
This is a revolutionary step forward for multimodal AI. For the first time, ChatGPT, with its o3 and o4-mini models, can integrate user-uploaded images directly into its thought process.
This chatbot not only identifies objects in images but also actually makes inferences based on those images.
For example, a user could send photos of several cars and ask for their names and model years, then go on to ask about their estimated value in 5 years—everything is quickly resolved in a single query.
Compared to previous inference models that often required waiting time, o3 and o4-mini offer significantly faster response times while maintaining depth and detail. The reports generated are extremely comprehensive, complete with clear source links.
The ability to automatically select and combine tools efficiently allows these new models to handle complex questions more smoothly, setting new standards for both intelligence and usefulness.
This is also seen as paving the way towards the promised flexible intelligent model transformation capabilities on ChatGPT-5 in the future.
Paid users (Plus, Pro, Team) can directly select o3, o4-mini, or o4-mini-high from the LLM menu in ChatGPT to experience the changes of the update.
Meanwhile, free users can activate o4-mini's reasoning ability by pressing the 'Reason' button before submitting a request (with a limited number of uses, much lower than paid plans).
Source: https://www.vietnamplus.vn/nhung-thay-doi-dang-chu-y-trong-ban-cap-nhat-cua-chatgpt-post1033736.vnp






Comment (0)