Months after updating ChatGPT’s image generation with Images 2.0, OpenAI on Tuesday released an upgraded model and a new Sketch tool that turns your doodles into high-quality AI-generated imagery.
Images 2.5 is currently rolling out to all ChatGPT users on desktop, mobile, and web. Compared with the older model, it reduces image-generation latency by 50% and is better at producing natural light and rich textures, preserving subjects from reference photos, and following edit instructions across multiple iterations, OpenAI says.
The model also adds new Templates to help speed up your creative work. Instead of starting from scratch, you can now go to Images > Templates and choose from pre-set formats, including Poster and Merch, to generate desired outputs quickly.
Another new tool sees ChatGPT turn your own rough drawings into detailed AI-generated images. Called Sketch, the feature opens a canvas with drawing tools. Just doodle what you have in mind, describe your vision in a short prompt, and the AI will finish the artwork.
You can get started by just typing @Sketch in the prompt box. I played around with the tool for a bit, and the results were quite impressive. Though it is being promoted as a sketch-to-AI Image tool, it appears that the prompt you pair with your sketch does much of the heavy lifting.
ChatGPT’s new image model arrives just days after OpenAI released GPT-6 Astra. It also introduced a new feature that lets you connect third-party messaging services directly to its chatbot, helping it better understand and emulate your writing style.
Disclosure: Ziff Davis, PCMag's parent company, filed a lawsuit against OpenAI in April 2025, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.