Nearly a week since OpenAI integrated its latest image-generation feature with the GPT-4o model, for users, it's been all about creating and reimagining life in the Ghibli-style artwork.
Such has been the scale of usage that OpenAI SEO Sam Altman has had to request people to give his GPUs a break. And now, Mr Altman has said the company is experiencing technical difficulties due to capacity challenges, adding they are actively working to resolve the issues.
The 39-year-old also said that OpenAI was working on updates but they were expected to be delayed.
"We are getting things under control, but you should expect new releases from OpenAI to be delayed, stuff to break, and for service to sometimes be slow as we deal with capacity challenges," he wrote on X. He also asked companies to help him with GPUs with a capacity to handle 1 lakh+ tasks at a time.
He wrote, "Working as fast we can to really get stuff humming; if anyone has GPU capacity in 100k chunks we can get asap please call!"
Soon, the comment section was flooded with people tagging companies to help him handle the situation.
One user said other features, such as the image generator, had also stopped working.
He wrote, "The image generator is not working as well now. The text generation is back to being mediocre. I'm a paid subscriber. What's the deal with that?"
Another commented, "Please just release things to Pro users with rate limits instead of outright delaying releases."
Someone wrote, "Honestly, that decision was kind of strange. Premium users are already having a hard time using the service because of server issues and super strict filters that block even simple prompts."
"And then free users were given access too, just adding to the overload with all their "make it look like Studio Ghibli" prompts. Honestly, I didn't really see any reasonable point behind that decision."
Last week, Mr Altman announced temporary rate limits, explaining the company's GPUs (graphics processing units) were "melting" under the pressure. He assured users that the team was working on making the tool more efficient.
On Sunday, he asked users to slow down on image generation, saying the demand was too high and the team needed sleep.
OpenAI calls GPT-4o its "most advanced image generator yet," offering improved accuracy, better text rendering, and more precise object relationships. Unlike DALL-E, which generates images in a single pass, GPT-4o constructs them step by step for enhanced realism.
