HiDream-I1: 17B Text to Image Generation Model for ComfyUI

HiDream-I1 is a 17B parameter text-to-image model combining DiT with MoE architecture. MIT licensed with Full/Dev/Fast variants.

H

HiDream-I1

Text-to-ImageOpen SourceDiTMoE

Open-source 17B parameter text-to-image model combining DiT (Diffusion Transformer) with MoE (Mixture of Experts) architecture by HiDream-ai. MIT licensed. Integrates four text encoders. Available in Full (50 inference steps), Dev (28 steps, distilled), and Fast (16 steps, most distilled) variants. Also available as fp8, GGUF (city96), and nf4 (azaneko) community packs.

Hugging FaceComfy-Org packGitHub
DeveloperHiDream-ai
Release Date2025-04-07
ArchitectureDiT + MoE (17B parameters)
LicenseMIT
Output Resolution1024x1024 native
VariantsFull (50 steps), Dev (28 steps), Fast (16 steps)
Community Variantsfp8, GGUF (city96), nf4 (azaneko)

Guides and workflows related to this model series.

No articles found.