Announcement from OpenAI Developers
GPT Image 1.5 is now available in the API:
More precise image editing and preservation of logos & faces
Better instruction following and adherence to prompts
Improved text rendering, particularly for denser and smaller text
This year’s “Little Shipmas” comes with a new image model.
Looking forward to what will be revealed later today.
Here’s a first post with images:
I have already updated the first post with these.
It handles photorealistic into abstract…
The old model struggled with combining such a thing….
(I bet Sam could get away with that style in Europe somewhere)
Quick clock test: it’s 9:30 not 11:55 
But that’s also the explanation why the clock is extra-large. I hope.
One-shot the crystals in a thread world I was shooting for back in July 
I fought for hours then and had to write huge contexts to do it…
Not anymore =D
Is this being rolled out in a staggered manner in the US? Hitting the models endpoint only shows gpt-image-1 available and trying to use gpt-image–1.5 with the v1/images/generation endpoint just results in:
Supported values are: 'gpt-image-1', 'gpt-image-1-mini', 'dall-e-2', and 'dall-e-3'.
I am getting
“error”:
{“message”:
“The resource you are requesting could not be found”,“timestamp”:
1765913734939,“name”:
“NodeApiError”,“description”:
“The model ‘gpt-image-1.5’ does not exist.”,
Doesn’t seem to be accessible to me yet either, but I assume it’ll be available soon.
My question (I didn’t notice directly addressed in the release) – has mask editing been improved?
Cheers!
me the same trying to use gpt-image-1.5 in the API i got a 404 “Not Found” ?
same issue here. just says “model not found”
It appears the model is still rolling out in the API.
Hang in there just a bit longer and sorry for making you wait! It’s definitely worth it.
Yeah, if its any consolation i keep getting:
An error occurred while processing your request. You can retry your request, or contact us through our help center at help.openai.com if the error persists. Please include the request ID req_09zdklasdlfaoipsdf in your message.
you can all test it by using an email and a free account in the chat
prompt:
An infografic explaining newton’s prism experiment in great detail
how many words is that?
niiiice.
I suspect the origin is this couch and a forum thread of putting people on it:
Aside from not being party colors - it looks like they broke the couch (which was prompted “sagging”) -and it grew in size!
Blog announcement link that first focuses on ChatGPT’s version:
https://openai.com/index/new-chatgpt-images-is-here/
It’s stronger at image preservation and editing than GPT Image 1.
You’ll see more consistent preservation of branded logos and key visuals across edits, making it well suited for marketing and brand work like graphics and logo creation, and for ecommerce teams generating full product image catalogs (variants, scenes, and angles) from a single-source image.
Image inputs and outputs are now 20% cheaper in GPT Image 1.5 as compared to GPT Image 1, so you can generate and iterate on more images with the same budget.
Let’s test it out as an “edits” model - can it outfill a mask with compatibility with the old image, and not alter unmasked areas (like DALL-E 2 could do)? What has been awaited for a long time?
By resize and reposition an input in my utility built for DALL-E 2:
The new model name is in effect.
Prompt: “Sliding penguin, with a mountain with a ski lodge in the background on a mountain.”
2 second animated GIF of before and after (transparency of original is shown):
Conclusion: your image input is still just part of a prompt to a multimodal model - all is regenerated.
Yeah… logos regenerating completely nullifies the logo…
But the claim was ‘more consistent preservation,’ which maybe we read differently.
My testing vector starts with the question now:
Does it handle logos differently when specified, or are they just referring to more consistent to reference input? I don’t think they actually claim not to regenerate…?
You’re always 5 levels ahead…
