HunyuanImage 2.1: 2K Text-to-Image Generation by Tencent

HunyuanImage 2.1 is Tencent's highly efficient 17B parameter text-to-image model capable of generating 2K (2048x2048) resolution images, repackaged.

H

HunyuanImage 2.1

Text-to-Image2K Resolution17B ParametersTencent Hunyuan

Tencent's highly efficient 17B parameter text-to-image model — generates 2K (2048x2048) resolution images. Features a two-stage architecture with a base diffusion transformer and a refiner model, dual text encoders (MLLM + character-aware), and meanflow distillation for efficient inference. Repackaged for ComfyUI with FP8 quantized variants.

Hugging FaceGitHubDemoPaper
DeveloperTencent Hunyuan Team
Release Date2025-09
ArchitectureSingle/Dual-Stream Diffusion Transformer (17B) + Refiner
Licensetencent-hunyuan-community
Output ResolutionUp to 2K (2048x2048)
Text EncodersQwen 2.5 VL 7B (MLLM) + ByT5 GlyphXL
VAE Compression32x spatial compression ratio
VariantsBase + Distilled + Refiner · BF16/FP8

Guides and workflows related to this model series.

No articles found.