Revealing why OpenAI stopped giving away free images

According to Digital Trends , OpenAI has been forced to impose temporary restrictions on the ability to generate images using its advanced AI model GPT-4o, after the trend of creating images in the style of Ghibli animations went viral online, putting huge pressure on the company's server system. Notably, this restriction also affects users who are paying for the service.

Ghibli craze overloads OpenAI servers

The cause of the overload was revealed humorously by CEO Sam Altman himself. He posted on social network X that while he was happy to see the online community constantly creating with the Ghibli art style, the surge in requests for image creation was at risk of 'melting GPUs' in the data centers, which can be understood as a metaphor for servers working at full capacity and possibly overheating.

CEO Sam Altman had to temporarily suspend the ability to create free images, overloading servers

PHOTO: SCREENSHOT X

In response, OpenAI announced that it would temporarily limit the speed at which images are generated while its engineers tweak the system to optimize performance. The company also acknowledged that ChatGPT may reject some image requests during this period and said it was working to fix the issue.

The move comes just a day after free users of ChatGPT were completely suspended from creating GPT-4o images due to overwhelming demand. While paid users are currently facing speed limits, free users will soon have access restored, though they will only be able to create up to three images per day, Altman said. A specific timeline for this has not been announced, however.

Despite the restriction notice, the online community seemed undeterred, continuing to 'attack' Altman's post with a series of other Ghibli-style images, showing the strong appeal of this trend.

Much of the appeal of the Ghibli movement comes from the impressive capabilities of GPT-4o. Not only does the model generate realistic images and handle text in images almost perfectly, it also excels at 'understanding' and reproducing the original images in a variety of artistic styles. Its ability to interpret context and reproduce with high accuracy in the signature Ghibli style makes GPT-4o a particular favorite for this trend, although other AI tools can do the same thing.

Source: https://thanhnien.vn/ven-man-ly-do-openai-ngung-cho-tao-anh-mien-phi-185250329093808534.htm