ChatGPT adds GIF creation to its image generation tools

New animation features

OpenAI has introduced a new version of its image generator called Images 2.6. This update allows users to create small looping animations, commonly known as GIFs, directly within the ChatGPT interface. The tool is designed to handle various types of requests, from entirely new creations to modifications of existing files.

According to numerama.com, the primary goal of this feature is to help users build content for social media. This includes creating custom memes, unique reactions, and animated mascots. Instead of needing external software to stitch frames together, the chatbot handles the motion logic internally based on user instructions.

The rollout began on November 4, 2026, and is being deployed across all ChatGPT and Codex plans. Because the feature is still being distributed globally, some users may not see the option immediately. OpenAI has confirmed that the tool is being phased in to ensure stability across different regions and accounts.

How to generate GIFs

Creating a GIF with the new tool is straightforward and relies on natural language prompts. Users can simply ask the chatbot to create an animation while specifying exactly what should move. For example, a user could request a GIF of a pink cat dancing, and the AI will interpret the motion and style from that single sentence.

The system also supports image-to-animation workflows. If a user uploads a static illustration, they can describe the desired movement to the chatbot. The AI first generates an image showing the character in various poses before producing the final looping animation. This process helps the model understand the physical range of the subject before it starts animating.

Beyond full creation, the tool offers some basic editing capabilities for existing files. Users can provide a link to an existing GIF and ask the AI to slow down the motion. The system also supports turning simple sketches into animations, allowing for a more tactile way to influence the final visual output.

Performance and details

OpenAI is not just adding new features but also optimizing the underlying technology of the image generator. Images 2.6 is designed to be significantly faster than previous versions. The company states that the update should accelerate the generation of images by 50% compared to earlier models.

In addition to speed, the update focuses on consistency and detail. One of the common challenges with AI image generation is losing specific details when a user asks for multiple revisions. This new version is built to better preserve those details throughout the editing process, ensuring that a character or object remains recognizable as it changes.

The tool also introduces a way to manage reference photos more efficiently. Users can now save a reference photo under a specific name and recall it in future conversations using an @ mention. This means a user could reference a specific pet or object without having to re-upload the file every time they want to generate a new GIF or image.

Content and limitations

The new tool has shown the ability to animate characters from popular media. In testing, the system successfully animated a character from the One Piece series despite it being a protected work. However, the results for these complex characters were noted to be less natural than simpler subjects like a basic dancing cat.

While the AI is capable of handling diverse prompts, the quality of the animation can vary based on the complexity of the source material. The system seems to perform most reliably when the subject is clearly defined and the requested motion is straightforward. Users may find that simpler subjects produce smoother, more fluid loops than complex scenes.

As the feature continues to roll out, OpenAI will likely refine how the model handles copyrighted material and complex physics. For now, the tool provides a versatile way for creators to experiment with motion. It serves as a bridge between static AI art and full-scale video generation, offering a middle ground for quick, shareable content.