Following its topping of the open-source image generation model arena list, HiDream.ai has achieved another milestone! Its latest open-source model, HiDream E1.1, has successfully entered the first tier of the Artificial Analysis list for image editing agents. As a leading open-source image editing model, it comprehensively outperforms mainstream models such as Flux.1 Kontext, continuously refreshing international standards with Chinese wisdom.

Released in July 2025 as an iterative masterpiece, HiDream E1.1 is built on the self-developed Sparse Diffusion Transformer architecture. It supports megapixel-level resolution and adheres to the MIT open-source license, delivering the revolutionary experience of “what you say is what you edit.” Users can complete complex tasks such as color adjustment, style transfer, and adding or removing elements through natural language, without the need for professional software operations. According to evaluation data from Hugging Face, the model has comprehensively surpassed its predecessor in terms of EmuEdit semantic understanding and ReasonEdit operation accuracy metrics, particularly demonstrating a qualitative advantage in the ability to execute combined multi-instructions.

Currently, the development version of ComfyUI has natively integrated HiDream E1.1 nodes, allowing developers to invoke it without modifying the code. This breakthrough marks the evolution of AI visual technology from the “image generation era” to the “semantic understanding + editing control era,” reshaping the boundaries of imagination in image editing through the power of open-source.

Contact

Company: HiDream.ai

Contact Person: Yuechong Zhai

Email: info@hidream.ai

Website: hidream.ai

Telephone: +86 13718564372

City: Beijing/Shanghai/Hefei