OpenAI has launched ChatGPT Images 2.5, a new version of the image generation model built into ChatGPT, the company confirmed on Tuesday. The update focuses on three persistent pain points of AI image tools: generation speed, the tendency of small edits to scramble entire images, and the difficulty of describing visual ideas in text.
According to OpenAI, Images 2.5 produces sharper details, more natural lighting and richer textures than its predecessor, and does a better job of preserving the likeness of people, pets and objects taken from reference photos. Generation is up to 50% faster than Images 2.0, the previous version. For more context on this story, see our ongoing latest AI developments.
Editing That Only Changes What You Ask
The most common complaint about AI image generators has been that requesting one small change — a different background, a corrected line of text — often redraws the whole picture. OpenAI says Images 2.5 is built to edit only what the user asks for while leaving the rest of the image untouched, even across multiple rounds of edits in a long conversation.
Axios, which received early access to the model, tested this by having the system design a tattoo, revise it from notes, and iterate on a logo through several rounds without losing the original concept — a multi-step workflow that typically breaks earlier image models.
Sketch, Templates and Comments
The update introduces a feature called Sketch, which lets users draw a rough doodle directly in ChatGPT — the layout of a room, the shape of an outfit — and use that drawing as the guide for the final image. Typing "@Sketch" in a conversation brings the tool up. OpenAI's bet is that pointing and drawing is often faster and more precise than writing another paragraph of prompt text.
Alongside Sketch, OpenAI added templates for common formats such as posters and merchandise, the ability to leave comments directly on an image to target specific edits, and an option to share the exact prompt behind a creation so others can remix it with their own photos.
Adele Li, OpenAI's product lead on images, told Axios the goal is to keep people actively involved in the creative process. "The sketch tool and these templates are kind of an attempt to make people more engaged in the process of creation, rather than them being a passive force," Li said.
Two New API Models for Developers
Developers are getting two new models through the OpenAI API. GPT-Image-2.5 Flare is aimed at high-volume, everyday generation such as social content and product photos, prioritizing lower latency. GPT-Image-2.5 Sunburst is positioned as the slower but more precise option for polished campaign work.
The split mirrors how OpenAI has tiered its text models, giving engineering teams a trade-off between cost and speed on one side and output quality on the other. Companies building image generation into their own products can now choose between the two without leaving the Images 2.5 family.
Watermarking and Provenance Stay In Place
OpenAI says the new models keep the safety checks used by previous image versions, along with C2PA content credentials and invisible watermarking to help identify AI-generated images. Provenance standards have become increasingly important as photorealistic generated imagery spreads across social platforms, and OpenAI has faced sustained pressure from regulators and rights holders to make synthetic media detectable.
A Competitive and Crowded Field
The launch lands in an increasingly crowded image generation market. TechRepublic notes that Canva has shipped more than 100 Visual Suite updates in 2026 targeting creative workflow friction, while rivals including Google and xAI continue to iterate on their own image models. Earlier this year, arena rankings showed OpenAI's GPT-Image-2 and xAI's Imagine trading places near the top of human preference leaderboards.
For OpenAI, the strategy with Images 2.5 is less about raw benchmark scores and more about interaction design: reducing the number of prompt rewrites needed to get a usable result. As TechRepublic put it, the update aims to make AI image generation feel less like typing a magic spell into a box and more like actually making something.
Availability
ChatGPT Images 2.5 is rolling out to all ChatGPT, ChatGPT Work and Codex users on desktop, mobile and web, according to OpenAI. The Flare and Sunburst API models are available to developers immediately, with pricing detailed on OpenAI's model documentation pages.
For small businesses and creators, the practical significance is straightforward: fewer regenerations to fix a label, a hand or a background means faster turnaround on the flyers, product shots and social posts where AI image tools are most heavily used. TechRepublic's assessment was that the update lowers the skill floor for getting a usable result, aimed squarely at people who don't want to become prompt-engineering experts just to produce a decent flyer or product shot. Whether Sketch and comment-based editing become the default way people direct image models — rather than long-form prompts — will be the metric to watch over the coming months.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
