Ultraman: OpenAI will release GPT-5 in the next few months, with capabilities better than expected.

On Friday local time, OpenAI CEO Altman announced that o3 and o4-mini will be released in the coming weeks, and GPT-5 will be released in the coming months. Altman posted on social media that the release of GPT-5 is later than expected, but for good reason. The most exciting news is that they now have the ability to make GPT-5 even better than originally anticipated. "We also found that integrating all the content smoothly is much more difficult than we expected. And we want to make sure we have enough capacity to support the unprecedented demand we anticipate." This indicates that OpenAI expects GPT-5 to be very popular after its release, with a large number of users likely to use it, so they need to be prepared in advance. Recently, with the new image generation feature of ChatGPT causing an internet frenzy, OpenAI's active user numbers, in-app subscription revenue, and download numbers have all hit record highs. It is reported that the number of paid users of ChatGPT has exceeded 20 million, compared to 15.5 million at the end of last year. The technical specifications of o3 and o4-mini have not been disclosed yet, but it is expected that they will bridge the capability gap between GPT-4 and the upcoming GPT-5. Industry observers believe that GPT-5 will have significant improvements in reasoning, planning, and memory functions. When OpenAI previously announced details about GPT-5, the company mentioned that they plan to offer unlimited chat access to GPT-5 for users in a "standard intelligence setting," but it will be limited by an "abuse threshold." ChatGPT Plus users will be able to run GPT-5 at a "higher intelligence level," while ChatGPT Pro users will be able to run GPT-5 at a "higher tier of intelligence." Earlier this year, Altman revealed, "GPT-5 will integrate features such as voice interaction, canvas, deep search, and deep research." He referred to a series of features that OpenAI has introduced in ChatGPT in recent months. He also stated, "One of our key goals is to unify our large models by creating a system that can use all of our tools to determine when to think deeply and when not to, and apply broadly to various tasks." Meanwhile, OpenAI is facing increasing pressure from competitors who are adopting an "open" approach to releasing models, such as China's AI lab DeepSeek. Unlike OpenAI's strategy, these "open" competitors provide their models to the AI community for experimentation and, in some cases, commercialization. However, Altman recently announced that OpenAI will release a "powerful new open-weight model with reasoning capabilities" in the coming months. This model will have reasoning capabilities and will undergo additional security evaluations. This article was originally published by "Cailianshe," GMTEight editor: Liu Jiayin.