OpenAI CEO Sam Altman has announced that the rapid adoption of ChatGPT’s new image-generation feature, introduced earlier this week, is straining the company’s servers, prompting a temporary restriction on its use.
Altman revealed on social media site X that the surge in image generation has significantly increased server load.
“It’s super fun seeing people love images in ChatGPT, but our GPUs are melting,” he said.
However, he confirmed that OpenAI will take steps to manage demand and prevent excessive strain on computing resources.
We are going to temporarily introduce some rate limits while we work on making it more efficient. Hopefully won’t be long! ChatGPT free tier will get 3 generations per day soon,” Altman wrote.
Backstory
Nairametrics reported that OpenAI recently integrated image generation directly into ChatGPT using GPT-4o, a move that eliminates the need for users to rely on DALL·E through separate platforms. The feature was rolled out on February 25 across Plus, Pro, Team, and Free subscription tiers.
- In a press release, OpenAI described this as a major step in making image creation an essential part of AI-driven communication. Altman also emphasized its significance, stating on X, “it’s an incredible technology/product.” He also noted that people are going to create some really amazing stuff and some stuff that may offend people.
- GPT-4o’s image generation includes enhanced capabilities such as text rendering, multi-turn generation, and instruction-following for better accuracy and creative flexibility. OpenAI noted that the model is particularly effective for generating clear text in images, refining visuals through conversation, and producing context-aware illustrations for educational and professional use.
The feature enables users to create high-quality visuals, including diagrams, infographics, logos, business cards, stock photos, and even customized digital art. It also supports editing professional headshots and generating personalized artwork based on existing images.
Viral adoption across social media
Since its launch, the feature has gone viral, with users sharing AI-generated anime-style renderings of their photos across X and other platforms. Altman himself updated his X profile picture to an image created using the new functionality.
The latest development builds on OpenAI’s previous success with its DALL-E model, one of the first AI-powered image generators introduced in 2021. In 2023, OpenAI integrated image generation directly into ChatGPT, further expanding its capabilities.
As part of the imposed restrictions, Altman announced that users on the free tier will soon be able to generate only three images per day. This measure aims to balance accessibility while managing the overwhelming demand on OpenAI’s computing infrastructure.