ChatGPT's 'GPUs Are Melting' As People Make Studio Ghibli-Esque AI Images

ChatGPT got a new artistic upgrade this week. Parent company OpenAI announced on Wednesday that it is bringing AI image generation directly to its flagship chatbot. And it's apparently so popular with ChatGPT users that CEO Sam Altman tweeted that the company's "GPUs are melting."

OpenAI's GPUs and servers probably aren't actually melting, but with heavier demand, it can use more energy (and heat up) and slow down requests. Altman said that in order to help manage the influx of requests, it's instituting temporary rate limits "while we work on making it more efficient." This isn't an unusual process for image generation -- typically, upgrading to paid plans can get you more guaranteed fast generations. It's kind of like how upgrading to a fast pass at Disney lets you cut long lines for rides. When ChatGPT's image generator tool launched yesterday, there weren't any of those different lanes.

In the few hours that ChatGPT's new image generator has been available, a trend has already emerged: People are using it to make images in the style of Studio Ghibli. In case you're unfamiliar: Ghibli movies like My Neighbor Totoro and Ponyo have an iconic art style. Altman even changed his profile picture on X to a Ghibli-ied version of himself. But tensions continue to rise between creators and AI companies. It's a revealing choice for OpenAI to lean into the style, especially, as Blood in the Machine tech writer Brian Merchant calls out, since its founder, Hayao Miyazaki, has said AI tools are "an insult to life itself."

Dall-E vs ChatGPT

In the past, if folks wanted to use ChatGPT to create images, they had to use Dall-E. Dall-E is a separate program that you can search for and use. It's our top pick for the best AI image generators, partly because its unique conversational style makes creating and editing images easy. Dall-E will still be available to use, and luckily, that "chat to edit" ability is included with ChatGPT's version.

You can use either Dall-E or ChatGPT no matter what plan you have. ChatGPT is available for free, with paid plans offering more features starting at $20 per month. If you're using ChatGPT, the limits of your current plan will apply to image generation -- if you're on the free plan, you may run into limits using the 4o model for messaging, file uploads and data analysis. Altman tweeted that free users will be limited to 3 generations per day soon but didn't specify when.

Using ChatGPT's image generation 

Image generation in ChatGPT 4o will focus more on creating work-related images, like infographics and diagrams. OpenAI says it's improved text rendering to make that happen -- something extremely necessary as AI consistently hallucinates and messes up words in images. You can also upload your own images and edit them with AI.

There are some serious limitations to ChatGPT's ability. Most importantly, it says that you may not be able to precisely edit specific regions of an image -- an essential task as AI models can hallucinate things like eleven-fingered hands. If you upload your own image and make edits to a subject's face, those changes may be lost from edit to edit. You may also see issues with cropping and struggles with data visualizations and multilingual text. The company says in a blog post that it is working on improving these things and hopes to introduce fixes as early as next week.

Like Dall-E, images made in ChatGPT don't have any visible watermarks denoting they are AI-generated. OpenAI said that its images will have C2PA⁠ metadata, an industry standard that lets folks know behind-the-scenes that an image is made by AI. In terms of safety, OpenAI says it will follow the same content guidelines as the rest of the 4o model. It said it has "heightened restrictions" around nudity and graphic violence.

How to create images with ChatGPT

It's easier than ever to bring your artistic visions to life with the help of AI. If you want to try your hand at ChatGPT's new image generator, here are a few ways to access the new feature.

• On the main page, click the three dots next to Search and Deep research. Click Create image.
• To make edits, simply type your request in the chat bar and send.
• When you're ready to use your image, click the down arrow in the upper right corner to download your file. 

In my initial, brief testing, I found that it was more likely to use correct information if you gave it a source to work with, like a URL to a primary source.

For more, check out our full review of ChatGPT and guide to writing the best AI image prompts.