ChatGPT Images 2.5 Is Here With Faster Generation and More Precise Editing
OpenAI says the new image model improves detail, editing consistency, and creative controls.
ChatGPT Images 2.5 is now rolling out, giving ChatGPT users a faster image-generation model with sharper details, more precise editing, and better consistency across multiple edits. OpenAI says generation latency is up to 50% lower than Images 2.0. OpenAI's announcement confirms the release and the new features.
The update is more than a model refresh. OpenAI is adding new controls around how people create and refine images, including a Sketch tool, templates for common formats, image comments, and the ability to share the prompt used to create an image.
What ChatGPT Images 2.5 changes
ChatGPT Images 2.5 improves image fidelity, with OpenAI saying it produces more natural lighting and richer textures while doing a better job preserving recognizable subjects from reference photos. It also follows focused editing instructions more reliably, so users can change one part of an image without unnecessarily altering the rest.
OpenAI also says the model is more consistent across multi-turn editing. Earlier changes are more likely to remain intact as users continue refining an image, which should make longer editing sessions less prone to unwanted visual drift.
The model is also better at handling complex visual instructions, including real-world information, detailed layouts, transparent backgrounds, and requested visual styles. These improvements are aimed at making more involved creative briefs easier to execute without repeatedly rebuilding an image.
Sketch, templates, and prompt sharing
Sketch gives users another way to explain what they want. Instead of describing an idea entirely with text, users can draw a rough layout or concept directly in ChatGPT and use it as a visual reference. OpenAI says the feature can be accessed by typing @Sketch in ChatGPT.
Templates are designed to reduce the blank-page problem. Users can start with formats such as posters or merchandise designs and then add the text, design elements, and style they want. ChatGPT can also ask follow-up questions to help refine a template-based request.
Image sharing now includes an option to share the prompt behind a generated image. That lets another person reuse the idea with their own photos and details instead of simply receiving the finished image.
ChatGPT Images 2.5 availability and API models
OpenAI says ChatGPT Images 2.5 is rolling out to ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web. The company's ChatGPT Images help page says ChatGPT Images is available on all tiers, while some image features may vary by plan or rollout.
For developers, OpenAI is also releasing two API models. GPT-Image-2.5 Flare is positioned as the default option for most applications, with the company claiming higher image quality than GPT-Image-2 and up to 50% lower latency. GPT-Image-2.5 Sunburst is aimed at workflows that need tighter control over detailed creative edits, with longer generation times.
The launch adds another major image-generation option to a field that is already moving quickly. Saganote has also covered Google's Nano Banana image generation in Google Earth and Meta's Muse image generator, while ChatGPT pricing in 2026 provides context on the plans through which ChatGPT features are offered.
What to expect from ChatGPT Images 2.5
For everyday ChatGPT users, the main change is faster iteration: create an image, make a targeted edit, and continue refining it with less waiting and fewer unintended changes. The new Sketch and template tools also give users more ways to communicate an idea when a text prompt alone is not enough.
OpenAI says the rollout is happening now across supported ChatGPT experiences, while the two API models are available for developers. The company's announcement also says Images 2.5 continues to use safeguards, C2PA metadata, and invisible watermarking for generated images. :::