Back in February, OpenAI CEO Sam Altman outlined the roadmap for GPT-5, stating that the powerful o3 reasoning model would not be released as a standalone version, but instead be integrated directly into GPT-5. However, there’s been a change in plans. Altman now says that both the o3 and the next-gen o4-mini reasoning models will be launching “in a couple of weeks,” with GPT-5 expected to follow “in a few months.”
On Friday, Altman posted on X, “change of plans: we are going to release o3 and o4-mini after all, probably in a couple of weeks, and then do GPT-5 in a few months.” OpenAI also plans to introduce an o3-pro model, similar to o1-pro, which will take more time to handle complex problems.
Altman has shared the reasons behind the delay of GPT-5, stating that integrating multiple tools into the model especially the o3 model has been “harder than we thought.” In the meantime, GPT-5 will continue to be refined, while OpenAI focuses on building the necessary infrastructure and capacity to support the model once it launches in a few months.
Previously, Altman confirmed that GPT-5 will be available to free ChatGPT users as well. Considering the recent surge in popularity of Studio Ghibli-style images on ChatGPT, it’s clear that OpenAI is wisely prioritizing capacity building.
Additionally, the early release of o3 may be strategically timed in response to Google’s flagship Gemini 2.5 Pro model, which has claimed the top spot in the AI industry. OpenAI may be aiming to reclaim leadership by introducing the powerful o3 model ahead of competitors like Google and Anthropic.
