OpenAI CEO Sam Altman has blogged about the temporary limits placed on ChatGPT’s new image-creating feature due to a deluge of usage overwhelming the firm’s servers.
Altman blogged the news on X (previously Twitter), hinting at the heavy demand:
“It’s super fun watching people love images in ChatGPT, but our GPUs are melting.”
To manage the crisis, OpenAI will implement limits on rates by which free-tiers can make only three images per day as the company continues to find an optimal way of managing the system.
OpenAI has integrated image generation itself into ChatGPT using the GPT-4o model recently, with a view to making it easy for users to create images without relying on various platforms like DALL·E.
It has been released on February 25 and is compatible with all Free, Plus, Pro, and Team plans.
It has been trending since its release, with users applying AI-created anime-style portraits, infographics, and other artwork extensively. It is even utilized by Sam Altman himself to modify his X profile picture.
The GPT-4o image generation feature operates to create highly detailed images, ranging from text-based graphics to context-sensitive images. Some of its most striking features are:
- Multi-turn generation (refining images through conversation)
- Accurate text rendering within images
- Personalizable digital images and infographics
But Altman also conceded concerns that the feature might be exploited to create objectionable or inflammatory material.
With demand skyrocketing, OpenAI is limiting free users to three image generations per day to prevent server overload and maintain stable performance.
Altman reassured users that the cap is temporary and that the company is working on making the system more efficient.
“Hopefully, it won’t be long!” he added.
