no test demo？
where can test？
You should either provide a test demo so we can validate the claims or clarify the following statement:
"HiDream-O1-Image (codename: Peanut) debuts at #8 in the Artificial Analysis Text to Image Arena, which is positioned to be the new leading open weights Text to Image model (2026-5-5)."
Otherwise, if the one listed on the arena is a closed pro version you will not be releasing and you're posting on your huggingface under the guise that it is that would be a pretty blatant lie. I hope the description can either clarify or be updated to reflect accuracy of what will be on the huggingface.
You should either provide a test demo so we can validate the claims or clarify the following statement:
"HiDream-O1-Image (codename: Peanut) debuts at #8 in the Artificial Analysis Text to Image Arena, which is positioned to be the new leading open weights Text to Image model (2026-5-5)."
Otherwise, if the one listed on the arena is a closed pro version you will not be releasing and you're posting on your huggingface under the guise that it is that would be a pretty blatant lie. I hope the description can either clarify or be updated to reflect accuracy of what will be on the huggingface.
Tested the quality of it and it's hugely deceiving. Can't even get text right like shown in the preview images.
at least you don't need a Nasa computer like for HYimage3. The text thing doesn't sound promising, even though the old Hidream was one of the first models to do it quite well. And could do other things that Flux.1 (at the time) just couldn't. So even though it never took off, i have good memories of it.
We'll see.
I have by the way a built in routine in my brain that just fades out any benchmark stuff. It's almost always a giant pr bs bubble.
They btw answered to a comment on the main model page that the spaces on hf will get some sort of update. New models are often a bit of a mess, so there's hope. Just wait for the "the NEW BEST INSANE local Image model" videos on YT :)
Thanks for your discussion. We have just released the whole pipeline of HiDream-O1-Image-Dev-2604 model on the Artificial Analysis Text to Image Arena in this link: https://huggingface.co/HiDream-ai/HiDream-O1-Image-Dev-2604
To be clear, HiDream-O1-Image and HiDream-O1-Image-Dev supports various tasks (e.g., text-to-image, image editing, and subject-driven personalization). In contrast,
HiDream-O1-Image-Dev-2604 (the whole pipeline with both prompt refiner and t2i model) is tailored for text-to-image generation task with higher image quality.
Thanks.