After effectively cancelling the consumer launch of its o3 reasoning model in February, OpenAI now says it aims to release both o3 and a next-gen successor, o4-mini, in “a couple of weeks.”
In a post on X on Friday, OpenAI CEO Sam Altman said the reversal in course is related to OpenAI’s upcoming GPT-5, which OpenAI has previously said will be a unified model incorporating so-called reasoning capabilities.
“[W]e are going to be able to make GPT-5 much better than we originally thought,” Altman wrote. “[W]e also found it harder than we thought it was going to be to smoothly integrate everything. [A]nd we want to make sure we have enough capacity to support what we expect to be unprecedented demand.”
Altman added that OpenAI expects to roll out GPT-5 “in a few months” — later than originally anticipated.
To the extent that OpenAI has published details about GPT-5, the company has said it intends to offer unlimited chat access to GPT-5 at the “standard intelligence setting” subject to “abuse thresholds.” ChatGPT Plus customers will be able to run GPT-5 at a “higher level of intelligence,” while subscribers to OpenAI’s ChatGPT Pro plan will be able to run GPT-5 at an “even higher level of intelligence.”
“[GPT-5] will incorporate voice, Canvas, search, deep research, and more,” Altman wrote in an X post early this year, referring to a range of features OpenAI has launched in ChatGPT over the past several months. “[A] top goal for us is to unify [our] models by creating systems that can use all our tools, know when to think for a long time or not, and generally be useful for a very wide range of tasks.”
OpenAI is facing increasing pressure from rivals such as Chinese AI lab DeepSeek that have adopted an “open” approach to launching models. In contrast to OpenAI’s strategy, these “open” competitors make their models available to the AI community for experimentation and, in some cases, commercialization.
In addition to o3, o4-mini, and GPT-5, OpenAI plans to debut its first open language model since GPT‑2 in the coming months. This model will have reasoning capabilities, Altman said earlier this week, and will be subject to additional safety evaluations.