The brand new [product] that OpenAI launched yesterday 4o Image Generation The feature was announced as available to free ChatGPT users as well, and many were eager to try it. But here’s the bad news: due to its overwhelming popularity after launch, usage appears to have been too high, leading OpenAI to postpone the rollout of the free version of 4o Image Generation. This means free users still cannot use it at this stage.
ChatGPT 4o image generation delayed for free users
This time, after OpenAI integrated image generation with GPT-4o, the generation capabilities have greatly increased compared to the previous DALL-E. Not only does it handle text better, but the effect of adding text to images is also quite impressive, and even Chinese characters are no problem.
In the previous introduction, we successfully had GPT-4o add specified text to images. In fact, the cover image of this article was also made by me using GPT-4o’s image generation, without any image editing software. The text portion was perfectly added by GPT-4o:
And the best part is that GPT-4o has already solved the image consistency problem. That means you don’t need to use local editing features—you can directly use text to ask GPT-4o to make partial modifications while keeping everything else identical to the original image. That’s really impressive.
Of course, that’s not all. Many netizens have also discovered other uses, such as changing image style, merging specified images into generated AI images, and multi-panel comics, among others.
And this will certainly cause a sudden surge in ChatGPT server load. Earlier, OpenAI CEO Sam Altman indirectly confirmed this and announced a delay in the release of the free version of 4o Image Generation.
ChatGPT 的圖像功能比我們原本預期的還要受歡迎——而我們當初的預期其實已經非常高了。因此，我們無法如期將圖像功能擴展到免費用戶，全面開放將會延後一段時間。
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
As for when it’s been postponed to? Sam Altman didn’t specifically say, and OpenAI’s official community didn’t mention any related information either.
Given the current popularity of 4o Image Generation, I guess we’ll have to wait a while. Although the first image generated by GPT-4o can already produce quite good results, it still needs a few fine-tuning tweaks to get the image you have in mind. For instance, the cover image for this post took me nearly 10 revisions.
Moreover, GPT-4o’s image generation speed is currently a bit slow—a single image takes tens of seconds to complete. If everyone uses it this way, the high load is bound to last for a while, until the hype gradually cools down or OpenAI adds new hardware.
Also, while using it, I noticed that after modifying the same image a certain number of times—say, more than 6 to 7 times—GPT-4o seems to get confused and generates a completely new image, requiring a new chat session to continue modifying it normally.
Some Chinese characters still cannot be generated correctly, such as the “圖” in “圖片” — it keeps generating the simplified character “图”. No matter how I try to change it, it won’t work.
Source: KOCPC Chinese