Is it qwen-image-edit？
same model arc and pram...
Thanks for your attention. FireRed Image Edit 1.0 is developed upon the Qwen-Image text-to-image (T2I) foundation model, having undergone a comprehensive training pipeline including Pre-training, Supervised Fine-Tuning (SFT), and Reinforcement Learning (RL) specifically optimized for image editing tasks. This training paradigm is backbone-agnostic and can be applied to other T2I models(currently based on Qwen-Image for better community support). For further details, please refer to the first paragraph of Section 3.1 in our technical report. The model demonstrates substantial performance gains in multi-image fusion and OOTD-related synthesis. Looking ahead, we will progressively open-source additional features, including our proprietary text-to-image (T2I) foundation model.
Therefore, FireRed Image Edit 1.0 shares the same text-to-image backbone (Qwen-Image, 20B parameters) as Qwen Edit 2509 and Qwen Edit 2511. However, it is not derived via post-training on the Qwen Image Edit model.
i'm not really convinced by it's editing jobs, slotting in (which you can 1:1 ) qe2511 gives a better (more consistent with the input image) result. BUT i quite like it as image generation model. It's definitely another option with different results to the qwen models.
Which of course you can do with an editing model, which can do so much more than just plain straightforward editing jobs. ("put this person in the following scene: completely different image generation scene" creating a new image, with that (roughly, hence the above) same person.