November 6, 2026, 5:22 pm | Read time: 2 minutes
A waving dog or a dancing cat: ChatGPT can now create short animations from your own photos. At the same time, the new image technology is designed to work faster and implement changes more precisely.
The new image function allows you to animate your own subjects. A photo or a hand-drawn sketch can serve as the starting material. In the chat, you then describe how the subject should move. ChatGPT creates a GIF file from this, a short animation that repeats continuously. This allows you to create personal greetings for family and friends without needing an additional image editing program.
Behind this is “Images 2.6,” the new version of ChatGPT’s tool for creating and editing images. Provider OpenAI introduced the update on November 4, 2026. According to the company, Images 2.6 is available to all users without a paid ChatGPT subscription. However, the GIF function is being rolled out gradually and may still be missing. The speed of created animations can be changed afterward. However, ChatGPT cannot edit existing and self-uploaded GIFs.
New images should be created significantly faster
OpenAI also promises improvements for static images. With Images 2.6, new images are supposed to be created 50 percent faster, according to the provider. This can be practical, for example, when trying out several drafts for an invitation and gradually adjusting colors, fonts, or subjects.
Frequently used photos can also be saved under a self-chosen name. Using “@Name,” ChatGPT can later recall the subject, even in a new chat. This means a portrait does not have to be re-uploaded each time if different variants or styles, such as a watercolor, are to be created from it.
Changes should better preserve other image details
OpenAI also claims to have improved post-editing. Change requests can still be described directly in the chat. The AI is supposed to more reliably retain details that are not meant to be changed. For example, if only the background of an image is replaced, the clothing of a depicted figure should remain as unchanged as possible.
Also of interest: Elvis songs used for AI training? Multimillion-dollar lawsuit against Anthropic
With this, OpenAI aims to reduce a known problem in AI image editing: A targeted change can currently lead to unintended changes in other elements of the subject. Images 2.6 is intended to make such additional corrections less necessary, working not only faster but also more controlled.