On Tuesday, OpenAI made its GPT-4o model-powered image-generation capability available to all users. Last month, ChatGPT’s paying users were the first to have access to the artificial intelligence (AI) image generating feature; today, even those on the free tier can use it. However, a high rate limit will be implemented, limiting the usage of free users. According to the San Francisco-based AI company, in order to reduce the number of deepfakes, each image created with this feature will incorporate Coalition for Content Provenance and Authenticity (C2PA) information into the metadata.
GPT-4o Image Generation Is Extended to Free Users by OpenAI
OpenAI CEO Sam Altman said in a post on X (previously Twitter) that ChatGPT’s GPT-4o-powered image-generation feature is now accessible to users on the platform’s free tier. Notably, in order to utilize the feature, users must be signed into their OpenAI account. Three image creation per day is the current rate limit for the free tier. There is also a cooldown time in between subsequent image generation.
Altman said in a different article that in only one hour on Monday, one million new users signed up for ChatGPT. He also contrasted it with the platform’s previous record, which was set when it began in 2022 and added one million users in five days.
The widespread Internet trend of people using ChatGPT to transform their photos into the aesthetic of Studio Ghibli, the well-known Japanese animation studio that created anime films like Spirited Away, Princess Mononoke, and My Neighbour Totoro, is one of the primary causes of the new feature’s popularity. People have been sharing their special moments on social media in the “Ghibli style” in droves.
Altman introduced temporary rate constraints for the premium ChatGPT tiers when employing the AI chatbot to generate photos because of how popular this trend was. The decision was made because “our GPUs are melting,” according to the CEO of OpenAI.
Separately, Altman recently attacked Meta, which now limits businesses with over 700 million monthly active users from using its open-source Llama models. “If your service has more than 700 million monthly active users, we won’t do anything silly like saying that you can’t use our open model,” he wrote in a post.
Notably, ChatGPT’s new picture generating capability includes prompt-based inline editing and better text rendering. Additionally, users can upload their own photos and request that AI alter or change them into different styles. According to OpenAI, the model rejects requests for photos that contain offensive material like violent or sexually suggestive deepfakes or images of child sexual abuse.
