OpenAI Launches ChatGPT Images 2.5
OpenAI has launched ChatGPT Images 2.5, its latest image-generation model. The update brings sharper details, faster generation and more precise editing.
The company announced the release on September 8, 2026. Moreover, OpenAI says Images 2.5 can produce more natural lighting and richer textures. It also preserves reference subjects more reliably during transformations.
Generation latency has also fallen by up to 50% compared with Images 2.0. As a result, users can create and refine images more quickly.
The update is available across ChatGPT, ChatGPT Work and Codex. In addition, users can access it through desktop, mobile and web platforms.
More Precise Image Editing
Images 2.5 improves how ChatGPT handles targeted edits. For example, users can change a product, background or piece of text while preserving surrounding details.
The model also handles complex subjects and backgrounds more consistently. It can also maintain earlier changes during longer editing sessions. As a result, users can make several revisions without rebuilding the image from scratch.
The model also improves reference-image fidelity. Consequently, people, pets and other subjects can retain more recognizable features across different settings and visual styles.
OpenAI has also improved its handling of complex layouts. The model can create images with real-world information and transparent backgrounds more accurately.
Sketch, Templates and Image Comments
Alongside the model upgrade, OpenAI has introduced several new creative tools. One of them is Sketch, which lets users draw directly in ChatGPT and use the drawing as a visual guide.
Therefore, users can sketch a rough room layout, clothing shape or other concept before describing the final image. ChatGPT then uses the sketch alongside the written instructions.
OpenAI has also added templates for common formats such as posters and merchandise. Users can select a template and then add their preferred information, design elements and style.
Meanwhile, users can place comments directly on generated images. This provides another way to specify targeted changes during editing.
Users can also share the prompts behind their images. As a result, others can reuse the same creative idea with their own photos and details.
New API Models and Availability
OpenAI is also bringing the Images 2.5 improvements to developers. The company has introduced GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst through its API.
GPT-Image-2.5 Flare provides the updated quality, editing and speed improvements. Meanwhile, Sunburst targets detailed creative work where users need greater control, although it takes longer to generate images.
OpenAI says Flare offers higher quality than GPT-Image-2 while delivering 50% lower latency. Therefore, the model is positioned for creator tools, product experiences, visual search and high-volume generation.
Safety remains part of the release. OpenAI says Images 2.5 uses checks on prompts and images to help prevent harmful outputs. It also continues to use C2PA metadata and invisible watermarking to identify images created with its tools.
For now, Images 2.5 is rolling out to all ChatGPT, ChatGPT Work and Codex users across desktop, mobile and web.
