AI Research Atlas

HunyuanVideo

Tencent · 3 December 2024

HunyuanVideo, a 13B-parameter open-source video model, matched closed models like Runway Gen-3 and Luma 1.6 in Tencent's 1,533-prompt human eval.

At release the largest open video model; released code and weights with a systematic report (data curation, architecture, scaling infrastructure). I2V (2025-03) and Avatar (2025-05) variants followed; HunyuanVideo-1.5 appeared 2025-11-18.

Date
Tuesday, 3 December 2024
Lab
Tencent
Kind
model
Access
open weights (restricted license)

Figures

MeasureValueMeasured by
Parameters13B+
largest open-source video model at release
company

Human-eval claims are Tencent's. Licence is Tencent's custom community licence (not Apache); regional restrictions apply.

Sources

  1. arxiv.org/abs/2412.03603
  2. github.com/Tencent-Hunyuan/HunyuanVideo

This record was checked against its sources on 6 October 2026. How we check