OpenAI launches Images 2.6 to create and edit images

OpenAI is launching Images 2.6, an update that creates and edits images faster, keeps details consistent and turns static designs into animated GIFs. It is now available in ChatGPT and Codex for all plans, on desktop, mobile and the web.

The tool is also coming to the API, the platform that lets developers integrate these capabilities into their own applications. Its goal is to reduce the time between a first idea and an image ready to use.

Images 50% faster

Images 2.6 generates images 50% faster than the previous version. In the API, developers can choose between two variants depending on the type of work:

• GPT-Image-2.6 Flare, designed for fast edits and tasks that need less detail. With quality set to low, it can generate an image in as little as six seconds.
• GPT-Image-2.6 Sunburst, designed for images with more detail, precision and consistency across different edits.

Flare also improves the composition of posters, product images and other professional designs. For example, you can change the text in an ad or the background of a product photo without disrupting the overall structure.

OpenAI says that 4 billion images are now created each week across ChatGPT, Codex and the API. That works out to more than 500 million per day, on average.

From a still image to a GIF

The main visible addition is the ability to animate images through GIFs. You can animate a brand mascot, create a reaction for social media or turn a simple sketch into an expressive character.

It can also help you prepare an animated storyboard before producing a video, test how a character moves or create a visual loop for a campaign. You do not need to start with a finished design: a basic drawing can become an animation.

The startup Town, which develops an AI assistant for work, used the API to explore how animated characters could make workplace interactions feel more personal.

More control for developers

The API adds several features that were already available in ChatGPT:

• Image marking to precisely indicate which area should be modified.
• SVG export, a format that lets you reuse designs on websites, applications and printed materials without losing quality.
• A complexity parameter to decide how much the model should plan before generating the image.
• Quality controls separate from complexity, so you can adjust the level of detail and the effort spent building the scene.

This separation lets you choose between speed and planning. For a simple edit, OpenAI recommends using low complexity. If the image includes many elements, constraints or specific instructions, it may make sense to raise it to medium or high.

For you, the most practical change is that creating an image no longer ends with the first result. You can iterate faster, point to exactly what you want to change and better preserve the identity of a character, product or brand across versions.

What to watch now is how these features perform in real commercial work: character consistency, text legibility, animation control and how easily you can take results from ChatGPT or the API into a campaign, website or application.