OpenAI CEO Sam Altman stated that ChatGPT is experiencing a surge in demand for its revamped image generation, prompting a temporary rate limit.

“It’s super fun seeing people love images in ChatGPT, but our GPUs are melting,” Altman confessed on X. That’s not just hyperbole.

The company had already delayed rolling out image gen to freeloaders, and now even paying punters are getting throttled—though Altman promises the choke won’t last forever.

The bottleneck emerges as OpenAI unleashes its upgraded GPT -4o model, producing visuals with enhanced realism and significantly improved text handling—an Achilles’ heel for AI art until now.

The firm told The Verge that it represents a “step change” from previous versions. No exaggeration there: what used to look like alphabet soup now resembles actual signage.

Free-tier users will soon receive a paltry three shots a day, which is enough to experience the feature but not use it.

Meanwhile, OpenAI scrambles to scale up infrastructure before the whole operation buckles under GPU shortages and rising energy demands.

The AI gold rush might be real, but it looks like the servers are paying the price.