OpenAI Images 2.6 With GIF Generation and 50% Faster Creation Rolls Out

• OpenAI Images 2.6 brings image generation to ChatGPT, Codex and the API, with creation up to 50% faster.
• The update adds animated GIF creation from text prompts, sketches and existing images for creative projects.
• Developers get two API models, image markup, SVG exports and separate controls for quality and planning effort.

OpenAI Images 2.6 is now available across ChatGPT, Codex and the OpenAI API. The new image generation system can create images up to 50% faster than the previous generation. It also adds a GIF creation feature that turns text prompts, sketches or existing images into short animated visuals.

The update is aimed at both everyday creative work and developer use. People can create animated mascots, product shots, social reactions, memes and storyboards before making a video. Developers can choose between two API models based on speed, image detail and consistency when making edits.

OpenAI Images 2.6 Adds Faster Image Creation and GIFs

OpenAI says its image tools now generate more than 4 billion images every week, which works out to over 500 million images a day. Images 2.6 builds on that usage with faster generation and a wider set of creative options. The speed improvement can help when users need to try several ideas or make changes to an image.

The main new feature is animated GIF generation. Users can describe the motion they want in a text prompt, provide a sketch or use an existing image as the starting point. The system can then turn that material into an animated GIF. This gives creators a way to test a visual idea without first moving into a full video editing process.

Possible uses include animating a brand mascot, adding movement to a product image, making a reaction GIF for social media or creating a short meme. Teams working on video projects can also use animated storyboards to preview scenes and movement before production begins.

Two Images 2.6 Models Are Available Through the API

Developers can access two models through the API: GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst. They are designed for different image generation needs, so the choice depends on whether speed or fine detail is more important for a task.

GPT-Image-2.6 Flare

Flare is the faster option. At low quality, image generation can take as little as six seconds. OpenAI also highlights its improved layout handling for work such as posters and product imagery. This model can suit tasks where quick results and clear placement of visual elements matter.

GPT-Image-2.6 Sunburst

Sunburst focuses on higher detail and better consistency across edits at medium and high quality settings. This can be useful when a project needs several versions of the same image while keeping important visual details consistent. For example, a product campaign may need different layouts or backgrounds without changing the product’s appearance.

New API Controls for Editing and Image Output

Images 2.6 adds several controls to give developers more direction over the result. Image markup lets a developer point to a specific area that needs an edit. This can make it easier to request a local change instead of describing the whole image again.

The API also supports SVG exports, which are useful for graphics that need to remain scalable. A new complexity parameter lets developers choose low, medium or high planning effort for a generation task. Image quality has its own separate control, allowing developers to set output quality independently of the planning effort.

These options provide more control over how the model works through a request. A simple image may need less planning, while a detailed poster or a multi-part composition may benefit from a higher planning setting. Developers can also choose the model that best fits the task rather than using one option for every type of image.

Safety Checks, GIF Safeguards and Image Watermarks

OpenAI says Sunburst uses multiple safety checks covering prompts, input images and generated outputs. ChatGPT image generation also adds synchronous classification for child sexual abuse material (CSAM). These checks are intended to help prevent harmful or prohibited image generation.

GIF generation has additional safeguards designed for risks linked to motion. OpenAI’s own evaluation reports 77.7% recall for these safeguards. That result does not mean every unsafe request will be detected, and the checks may not catch everything.

Images generated with the tools carry C2PA metadata and an invisible SynthID watermark. These technologies add information that can help identify AI-generated content. They do not change the fact that users should consider the context and accuracy of an image before sharing it as real-world material.

OpenAI Images 2.6 Availability

OpenAI says the new image generation features are rolling out globally to all ChatGPT subscription tiers. They are available on the web, desktop and mobile apps, and through Codex. Developers can access both GPT-Image-2.6 models through the API.

OpenAI also shared examples of how businesses are using the technology. Startup Town used the Images 2.6 API to create animated characters. Figma Weave’s creative director highlighted the speed and control available with the updated tools. These examples show how image generation can fit into product design and other creative workflows.

The release adds to OpenAI’s wider set of ChatGPT tools and features. For more updates, see our coverage of OpenAI’s Dots updates and the changes to ChatGPT usage limits.

For the full feature announcement, visit the official OpenAI Images 2.6 announcement.

Conclusion

OpenAI Images 2.6 brings faster image creation and text- or image-based GIF generation to ChatGPT, Codex and the API. The two API models offer a choice between speed and higher detail, while markup, SVG export and new controls give developers more ways to guide image output. Safety checks and invisible watermarking are also part of the release.

Frequently Asked Questions