Tencent releases Hunyuan Hy4-flash on Hugging Face: MoE multimodal image-text-to-text model

tencent · hf · 2026-11-17

Tencent has released a new model, tencent/Hy4-flash, on Hugging Face under the Hunyuan family. It ships with transformers and safetensors support, uses a Mixture-of-Experts architecture, and runs an image-text-to-text pipeline for instruction-following multimodal generation.

Related event: Tencent open-sources Hunyuan Hy4-flash with 1M context(4 posts)→

Original post →

More from Models

Models channel →