OpenAI has recently added GPT-4o image generation capabilities to ChatGPT, allowing users to create images directly within the app. The company claims that its new model will be able to tackle the usual shortcomings of other AI image-generators on the market, by being able to render texts and follow prompts accurately. Users are also able to modify uploaded images or use them as a visual inspiration to generate new ones.
The company says GPT-4o uses an “autoregressive approach” to its image generation, so the pictures will be produced from left to right and top to bottom, instead of all at once. Users are able to access the model directly in ChatGPT as its default image generator, and the images can be customised by specifying its aspect ratio, colours using hex codes, or with a transparent background.
The model does come with its limitations, such as needing a longer time to generate more detailed pictures. At its launch, it is also prone to cropping longer images too tightly at the bottom, being inaccurate with certain information rendering and “hallucinating” images with more complex requests, among others. However, OpenAI states that these flaws will be fixed within the upcoming months.
images in chatgpt are wayyyy more popular than we expected (and we had pretty high expectations).
rollout to our free tier is unfortunately going to be delayed for awhile.
— Sam Altman (@sama) March 26, 2025
While the model was rolled out to all ChatGPT Plus, Pro and Team subscribers, OpenAI announced free users will only get it much later. Through a post on X, company CEO Sam Altman stated that the delay was caused by the model being “wayyyy more popular” than what was expected. There has been no further updates as to when the new model will become available to those not subscribed to ChatGPT’s premium plans, however.
(Sources: OpenAI, via Neowin / The Verge)
Manisha Dharmendra contributed to this article.
