OpenAI unveils ChatGPT Images 2.5 with 50% faster generation and new Sketch feature
OpenAI has launched ChatGPT Images 2.5, a new image model that improves fidelity, editing precision, and multi-turn consistency. The update cuts generation latency by up to 50% compared to Images 2.0. New features in ChatGPT include Sketch, which lets users draw directly in the chat as a reference, and templates for common formats. In the API, two models are introduced: GPT-Image-2.5 Flare for speed and quality, and GPT-Image-2.5 Sunburst for premium workflows. The model is available across all tiers of ChatGPT, ChatGPT Work, and Codex.
Every week, people create more than 3 billion images across ChatGPT Images and the GPT-Image models in the API.
- jjcm
I use gpt image 2 very heavily for my current project (ai UI design tool). The biggest improvement I'm seeing with this is in speed. I've generated around 50k images with gpt-image-2 via api, the the average latency has held at around 104s.
It's wild how much of a difference this is - images are coming in at around 35-40s. Very noticable, and makes a difference when you're iterating quickly: https://jjcm.org/gpt-image-2.5-speed.mp4
- meerita
I've really enjoyed using AI to generate images. For example, Long time ago, I read an amazing five-book saga called Riverworld, and I used AI to recreate many scenes, places, and ideas from the story. Seeing the books come to life through hundreds of images was a great experience. Definitely one of the coolest things to enjoy in 2026. Try it with your favorite books.
- AyanamiKaine
Omg, I love how the first examples just show how easy you can fake things. Fake being at a party with your friends. Didn't make your bed, no problem, just fake it.
The sad part is my mother would love "remixing" my old child photos of me.
- pelzatessa
The "composite party photo", while impressive, shows that still the miniscule details are being lost, like the teeth structure of the guy in the middle or the fact that the guy on the left is holding the cup with three fingers. Wondering why they chose this edit for the showcase.
- minimaxir
A big wtf at the LM Arena scores: https://arena.ai/leaderboard/text-to-image
gpt-image-2.5-sunburst: 1421
gpt-image-2.5-flare: 1399
gpt-image-2 (medium): 1381
mai-image-2.6: 1331
Even with LM Arena being flawed, this is significant. I was planning to do a writeup on the original gpt-image-2 as it crushed every complex image comprehension benchmark I had...I'm glad I procrastinated since ChatGPT Images 2.5 seems like an even better starting point to test out what these models can actually do nowadays.
Many people still think AI images output the wrong number of fingers on a regular basis. (EDIT: this was an ironic comment to make in hindsight and I own it)
- tyjen
The Facebook Marketplace experience for used items has depreciated considerably with the widespread adoption of LLMs. Between placing attractive models in photos to help sell items to "improving" the visual condition of items that completely misrepresent the real condition of the item, it's becoming a trickier landscape to navigate to find what you're looking for.
- vgalin
The introductory video is a bit... daring. Generating tattoos that clearly look AI-generated is not one of the use-cases I'd try to sell.
- Nition
Did anyone else notice that all the use cases they show here are "use ChatGPT Images to imagine all the things you'd like to have, but don't"? For some, they then show the thing being done, since just imagining it is a bit sad. But ChatGPT Images can't help you with that part. At best, it's providing creative inspiration - but that's often the most rewarding part of the process for a human... None of the examples show things where you just need an image directly, e.g. product labels, signage, sprites, textures, images for websites etc. I guess aspirational stuff sells?
In order, there's:
- Imagine you had better decorations for your fish.
- Imagine you had a tattoo of your pet.
- Imagine you had a unique candle holder.
- Imagine you had a worse haircut.
- Imagine your flowers were arranged.
- Imagine your child had a suit.
- Imagine your dog had a costume.
- Imagine you owned a scanner.
- Imagine you could hang out with friends.
- Imagine your bed was made.
- bufbupa
Still can't make sprite sheets :(
- tezza
I've done a side by side comparison of the two new models across all quality levels. Versus all the previous models
https://generative-ai.review/2026/09/rush-openai-image-gen-2...
right, off to bed now