---
title: "Tencent opens the weights of Hy4 flash"
date: 2026-11-18
updated: 2026-11-18T00:00:00Z
url: https://arun477.github.io/feed/2026-11-18/
author: Arun Arumugam
---

# Tencent opens the weights of Hy4 flash

Tencent put Hy4 flash, a 224B MoE with 12B active and 1M context, on Hugging Face as open weights, along with FP8 and base builds.


- [Tencent opens the weights of Hy4 flash, a 224B MoE with 12B active and 1M context](https://x.com/TencentHunyuan/status/2122912575302119554) (Tencent Hunyuan, release, Models): Two days after the cheap API launch, Tencent put Hy4 flash, an FP8 build and the base model on Hugging Face. A natively multimodal 1M-context model you can self-host is a strong new option for open-weight agent work, though 224B total parameters still needs a multi-GPU node.
