 |
GPT-4.1, OpenAI's flagship model. Photo: The Verge . |
According to The Verge , OpenAI plans to release a series of new AI models. One of them is GPT-4.1, an improved version of the previous multimodal GPT-4o. It was originally introduced in 2024 as a flagship product, capable of processing audio, images, and text simultaneously in real time. According to some sources, OpenAI will soon announce GPT-4.1 along with smaller versions, GPT-4.1 mini and nano, as soon as next week.
In addition, the company is also launching at the same time or earlier, a full version of the o3 inference model and a reduced version o4 mini. AI engineer Tibor Blaho discovered references to o4 mini, o4 mini high, and o3 in a new web version of ChatGPT on the morning of April 10, suggesting that these additions are coming soon.
At the same time, OpenAI CEO Sam Altman also hinted at the launch of the new model on X. However, some sources said that OpenAI recently delayed the introduction of the model due to processing capacity issues and user overload.
Recently, OpenAI on X posted about ChatGPT’s new feature, which allows the chatbot to tailor its responses based on the content of previous conversations. This means that the AI remembers everything the customer has told the chatbot, including preferences and priorities. “The AI system can learn about you over time, becoming extremely useful and personalized,” CEO Sam Altman wrote on X.
 |
You'll know your account has been upgraded with smarter storage when you see a splash screen like this. Photo: OpenAI. |
The new memory feature will be rolled out first to ChatGPT Pro and Plus users. If you don't want to be remembered, you can turn the feature off in settings, delete saved memories, or switch to Temporary Chat mode.
According to Techradar , this feature is not too new because OpenAI has been able to remember user information before. From 2024, you will be able to ask the chatbot to forget or save specific details.
The difference of the new version is that the chatbot will automatically remember important information without the customer having to say it explicitly, helping to create contextual conversations. In addition to text, it also applies to voice and images.
However, according to Techradar's hypothesis, no new models will be released by OpenAI in the near future. The article reveals that Sam Altman's new model may only serve as an advertisement for this new feature.
While not entirely new, the chatbot's storage capabilities are a useful solution for users. In addition, Sam also committed that this new update will improve the speed and performance limits of ChatGPT.
Source: https://znews.vn/openai-se-ra-mat-mo-hinh-moi-post1544965.html
Comment (0)