OpenAI launches Images 2.6 with GIF animations and SVG export

New motion and export tools

OpenAI has released Images 2.6, a significant update that introduces motion capabilities to its image generation suite. Users can now transform static elements, such as a product shot or a character sketch, into looping animations. This feature is designed to help creators build social media reactions and test storyboards before committing to full video production.

Beyond animation, the update adds SVG export functionality. Scalable Vector Graphics (SVG) are files that can be resized without losing quality, making them essential for professional use in web design, mobile applications, and print media. This shift suggests a move toward making AI-generated content more practical for integrated brand workflows.

The release is available across all subscription tiers and platforms, including ChatGPT, Codex, and the OpenAI API. While the company describes the launch as global, it does not provide specific details regarding unique features for any individual region. The update represents the latest iteration since the release of Images 2.0 in April.

Two models for different needs

Developers now have access to two distinct models within the Images 2.6 framework, each optimized for different priorities. The GPT-Image-2.6 Flare model is built for speed, with OpenAI claiming it can return an image in as little as six seconds when set to low quality. Flare is also intended to improve the layout of posters and product imagery.

For tasks requiring higher fidelity, the Sunburst model is the primary option. It is pitched as providing higher quality and better consistency between different edits when using medium or high quality settings. This distinction allows users to choose between rapid prototyping and polished, repeatable results for complex projects.

To manage these models, developers can now use a complexity setting to determine how much the AI plans its output. OpenAI recommends using the low complexity setting for most standard edits. This level of control is intended to help teams maintain a steady output while balancing the computational demands of the generation process.

Technical updates and API changes

The update also involves a shift in how specific image modifications are handled. Image markup, a tool that allows users to mark exactly what should change in a generated image, has moved from the ChatGPT interface to the API. This change is primarily aimed at teams who need to automate or integrate precise editing into their own software stacks.

OpenAI also reported that the system handles a massive scale of production, claiming to generate over half a billion images daily. These figures, provided solely by OpenAI, underscore the high volume of traffic the new infrastructure is expected to support. The update also includes the inclusion of C2PA metadata and SynthID watermarks on all outputs.

Safety testing was also conducted as part of the release. OpenAI's system card reported tests on 3,077 adversarial requests to see how many unsafe images bypassed moderation. While the results showed some fluctuations in the percentage of safe outputs for both Flare and Sunburst, the company stated that these changes were not statistically significant.

Impact on marketing and design

For marketing and design teams, the practical impact of Images 2.6 centers on brand consistency and faster iteration. The ability to create consistent assets across multiple edits means that companies can more easily maintain a unified visual identity. This is particularly useful for creating a series of marketing materials that need to look like they belong to the same campaign.

The addition of GIFs provides a low-friction way to create motion content. Because these are looping animations rather than full videos, they can be quickly produced for social media. However, the speed of production means that marketing and legal teams will need to review more material to ensure it complies with existing advertising and intellectual property rules.

Because the pace of these updates is so rapid, workflows that were approved in the spring may already be outdated. Organizations should consider testing these new features on low-risk assets first. This allows teams to understand the limits of the motion and consistency tools before deploying them on major brand campaigns.

Verification and next steps

While OpenAI provides several performance claims, such as the 50% increase in generation speed, these figures are based on the company's own internal testing. Independent testing will be necessary to confirm if these speeds and consistency levels hold up in real-world production environments. Users will need to evaluate how the models perform under specific, high-pressure tasks.

The release also highlights a growing complexity in AI safety and transparency. The system card lists training data only in broad categories and does not address specific rights regarding the output. This means that firms should not interpret the current documentation as a final settlement on the legalities of AI-generated content.

Moving forward, the focus for users will be on mastering the new complexity settings and choosing the right model for the right task. Whether a project requires the quick turnaround of Flare or the high-quality consistency of Sunburst will be a key decision for developers. The transition toward more production-ready files like SVGs marks a significant step in the utility of these tools.