Popular ChatGPT feature "melts" OpenAI GPUs

Due to the enormous interest in the new image generation feature, OpenAI has faced an unexpectedly high load on its servers. As a result, the company is temporarily limiting the number of generation requests for all users and postponing the launch of this feature for free accounts.

It was announced by OpenAI CEO Sam Altman on his X page.

GPUs are melting because of the influx of users

According to Sam Altman, the demand for the new ChatGPT option has been so great that GPUs are melting down from the excessive number of requests. The company is introducing temporary speed limits to improve performance and prevent the servers from being completely overloaded.

"It's super fun seeing people love images in ChatGPT, but our GPUs are melting. We are going to temporarily introduce some rate limits while we work on making it more efficient. Hopefully won't be long," Altman noted in his X account.

He also added that free ChatGPT users will soon be able to create three images per day.

ChatGPT's image creation feature is based on the GPT-4o model (not DALL-E) and is capable of generating more realistic and detailed images than previous versions. Other innovations include better text rendering quality and direct integration with chatbots. Users note that the results of this model exceed the capabilities of competitive systems.

Many memes and examples of different styles of generated images have already appeared on the web. The service was overloaded after the emergence of a new trend for generating images in the style of Ghibli cartoons.

OpenAI has not yet named specific dates when image generation will become available to everyone without exception. At the moment, it is available only to those who have a paid ChatGPT subscription. Those who use the free version are advised to wait for an announcement from the company. The developers promise that after optimisation, they will be able to generate several images daily.

As a reminder, OpenAI has updated the Advanced Voice feature, which provides voice communication with ChatGPT. The company aims to make the chatbot more friendly and get rid of its annoying habit.

We also wrote that OpenAI released its most powerful AI version o1-pro in the API for developers. However, the cost of using the model turned out to be too expensive.