The AI startup OpenAI has incorporated an image generator based on the GPT-4o model into ChatGPT and Sora. This feature is available to all users of the company’s products.
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
Previously, the DALL-E 3 model was used for image creation in ChatGPT, while GPT-4o handled text. Now, the latter is also employed for generating images in the chatbot.
GPT-4o “thinks” a bit longer than DALL-E 3 during the creation process. This is necessary for generating more accurate and detailed images, OpenAI emphasized. The model can edit existing pictures, including those with people, transforming or removing details—objects in the foreground and background.
Creating and customizing images is as simple as chatting using GPT‑4o—just describe what you need, including any specifics like aspect ratio, exact colors using hex codes, or a transparent background. pic.twitter.com/fnen1aDjNZ
— OpenAI (@OpenAI) March 25, 2025
“Creating and customizing images is as simple as chatting using GPT-4—just describe what you need, including any specifics like aspect ratio, exact colors using hex codes, or a transparent background,” OpenAI emphasized.
The startup’s CEO, Sam Altman, highlighted the “incredibility” of the new product.
we are launching a new thing today—images in chatgpt!
two things to say about it:
1. it’s an incredible technology/product. i remember seeing some of the first images come out of this model and having a hard time they were really made by AI. we think people will love it, and we…
— Sam Altman (@sama) March 25, 2025
“I remember seeing some of the first images made by this model and couldn’t believe they were really created by AI. We think people will love it and eagerly await the results of creative activity,” he wrote.
He showcased an image generated during the presentation of the new tool. Users noted that the AI still hasn’t learned to create five fingers for people.
bro got violated pic.twitter.com/TMA9luo3id
— adi (@adonis_singh) March 25, 2025
The company emphasized the presence of censorship to combat the generation of images that may violate the firm’s policy. This includes materials on child sexual abuse, fakes, nudity of real people, and so on.
In March, Sora’s head of development, Rohan Sahay, заявил OpenAI’s intention to integrate a video generator into ChatGPT.
In December 2024, the startup выпустил a tool for creating videos to the public. The neural network creates clips based on text prompts, “animates” images, expands existing works, and fills in missing frames.