OpenAI
APIs and tools for building AI products
5.0•893 reviews•41K followers
APIs and tools for building AI products
5.0•893 reviews•41K followers
The most powerful platform for building AI products.
Build and scale AI experiences powered by industry-leading models and tools.
This is the 56th launch from OpenAI. View more
ChatGPT Images 2.6
Launched this week
Images 2.6 brings faster image generation, more precise editing, and native GIF creation to ChatGPT, Codex, and the API. Create looping animations, maintain character and product consistency across edits, export SVGs, and control generation complexity.
Free Options
Launch Team
Render Sandboxes: a computer for every agentRun untrusted code safely. Idle sandboxes cost nothing
Promoted
Gemini 2.5 Pro
Claude Code
Images 2.6 makes it easier to go from an image idea to a finished visual. Generate images faster, make targeted edits, and even turn static images into animated GIFs.
Key features:
Faster generation: Create images faster, with Flare generating images in as little as six seconds at low quality.
GIF creation: Turn prompts, sketches, and existing images into looping animations.
Better consistency: Preserve character appearances, products, and brand details across multiple edits.
Two API models: Choose Flare for speed or Sunburst for higher-quality, detailed visual work.
Image markup: Point to specific areas of an image and describe exactly what needs to change.
SVG export: Generate scalable vector graphics for websites, apps, and print.
Complexity controls: Adjust how much planning the model does, separately from image quality.
Who it's for: Designers, marketers, content creators, developers, and businesses creating visual content at scale.
Try Images 2.6 in ChatGPT | API
P.S. I hunt the latest and greatest launches in tech, SaaS and AI, follow to be notified → @rohanrecommends
I haven't tried with 2.6 yet, but I've found GPT Image has taken a few really big steps back in terms of Photorealism lately, has anyone else been finding this? Even running API tests against sunburst with 2.5, I was struggling to get stuff that didn't look a bit CG-ish. I have to assume that this is a guard rail rather than intentional.
Has anyone else been finding the same thing? And is this solved in 2.6?
The SVG export is the part that caught my eye. I build SwiftUI apps, so being able to generate an asset and then point at a specific area to tweak it — rather than re-prompting the whole image — would save a lot of round-tripping. Curious how clean the exported paths actually are, and whether layers stay editable once it's out as an SVG?
For SVG you can choose how to split the parts of let's say a character?
Turning static images into looping GIFs is a pretty cool addition. Curious to see how well it works in practice!