A new image generation artificial intelligence (AI) model recently released by OpenAI is experiencing high popularity, leading to server overload. In response to the surge in users, the use of this feature will be temporarily limited.
Sam Altman, OpenAI's Chief Executive Officer (CEO), noted on social media platform X on the 27th (local time), "It's really enjoyable to see people liking the image model, but the GPUs are melting down," referring to the server burden due to excessive usage. He added, "We plan to limit the use of the feature until performance improvements are made."
On the 25th, OpenAI launched the multimodal image generation model "ChatGPT-4o Image Generation," which can understand and generate text and images. This model is characterized by its ability to more accurately grasp user intentions and significantly improve text insertion accuracy compared to previous versions.
For example, if a user requests a "poster showing various types of whales," the model can accurately match each whale's type and name in the generated image and can be used for various forms of creative work, including diagrams, infographics, logos, business cards, and pet drawings.
The feature is available to both paid subscribers of ChatGPT and free users, but free users are limited to a maximum of 3 images generated per day.