OpenAI CEO Sam Altman has announced a delay in the launch of the much-anticipated GPT-5 model. The decision comes after an overwhelming response to the latest update in ChatGPT’s capabilities—its ability to generate images directly using the new GPT-4o model.
Last week, OpenAI rolled out a major update to GPT-4o that introduced native image generation. This means the chatbot can now create images on its own without relying on external models like DALL-E 3. The update significantly improved the chatbot’s understanding of user requests, allowing it to produce more accurate and creative images. One of the most popular trends that emerged from this update was the creation of Studio Ghibli-style artwork, which quickly went viral on social media platforms.
OpenAI Delays GPT-5 Launch Due to Massive Demand for Image Generation
However, this unexpected popularity came at a cost. The surge in image-generation requests caused a shortage of graphics processing units (GPUs) within the company. Even paid users, including ChatGPT Plus and Pro subscribers, faced limitations on how many images they could generate. In response to the situation, OpenAI acknowledged the need to better prepare for high demand in the future. “We want to make sure we have enough capacity to support what we expect to be unprecedented demand,” the company stated.
In a post on X (formerly Twitter), Altman confirmed that he will launch the GPT-5 a few months later than initially expected. He mentioned that the delay was due to several reasons, with the most exciting one being the opportunity to make GPT-5 significantly better than originally planned.
“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,” Altman posted.
What Do We Know About GPT-5?
So far, OpenAI has kept most details about GPT-5 under wraps. However, Altman previously mentioned that GPT-4.5, which was released recently, would be the last model that doesn’t focus on reasoning. The goal for GPT-5 is to merge the capabilities of the “o” series (such as o1 and o3) with the GPT series to create a unified system. This system would be smart enough to choose whether it needs to think deeply or quickly, depending on the user’s request.
OpenAI also plans to simplify its growing list of products and models. Altman stated that GPT-5 will offer “unlimited access” at standard intelligence levels (with some safeguards), and higher intelligence settings will be available for Plus and Pro users.
New Models Coming Soon
Before GPT-5, OpenAI plans to release three new models—o3, o3 Pro, and o4 Mini. The o3 and o4 Mini models will be available in the coming weeks. These models belong to OpenAI’s newer “reasoning-first” category, designed to better handle complex tasks like programming, mathematics, and logical problem-solving.
In addition, OpenAI is also preparing to release its first open weights model since GPT-2. This move will promote transparency and collaboration but will undergo extra safety checks before becoming available to the public.
