OpenAI Launches 4o Image Generation To Produce More Realistic Images

OpenAI launched its most advanced image generation tool into its GPT-4o model this Tuesday. The company claims the new o4 image generation can produce more realistic and useful images, and performs better than previous DALL·E 3 models. The new AI tool has started to roll out to all users globally, including free accounts..

According to the information shared by OpenAI on its website, the new tool can create infographics, simulate paintings, create comic strips, and generate photorealistic images. The new tool also includes features like processing images for inspiration and handling up to 20 objects per image.

“GPT‑4o image generation excels at accurately rendering text, precisely following prompts, and leveraging 4o’s inherent knowledge base and chat context—including transforming uploaded images or using them as visual inspiration,” states the document.

The company shared multiple images—on its website and on social media—to showcase the image’s capacities and the results it generated from each prompt.

4o image generation has arrived.

It’s beginning to roll out today in ChatGPT and Sora to all Plus, Pro, Team, and Free users. pic.twitter.com/pFXDzKhh2t

— OpenAI (@OpenAI) March 25, 2025

OpenAI also acknowledged the AI tool’s limitations and assured that its team will optimize the tool after its launch phase. The company also addressed the risks of the use of this advanced tool, assuring that they will block any request for sexual content and that all images will contain C2PA⁠ metadata—the information to track and verify their origin.

“When images of real people are in context, we have heightened restrictions regarding what kind of imagery can be created, with particularly robust safeguards around nudity and graphic violence,” said OpenAI.

While advanced tools to generate realistic AI images can provide multiple benefits and good uses, as the ones mentioned by OpenAI,— video game development, advertising, and creative experiences—the risks related to the use of these advanced technologies are also evolving.

Malicious actors across the globe have already been relying on other realistic AI image generators for deepfake scams and extortion. Google recently reported to the Australian government over 250 complaints of the use of Gemini for terrorism and child abuse material. In February, the United Kingdom became the first country to criminalize AI-generated child abuse tools.