ApX logoApX logo

HunyuanVideo 1.5 720P T2V

Parameters

13B

Modality

Video

Auxiliary Parameters

6.36B

Architecture

Diffusion Transformer

License

Tencent Hunyuan Community License

Release Date

18 Feb 2025

API Pricing (per 1M)

Self-hosted only

Hardware Sizing

Precision

Workload Memory: 8.98 GB

INT4 · 1024x1024

DiT Backbone Weights: 6.05 GB

VAE Decode Workspace: 0.42 GB

Runtime & Graphs: 2.50 GB

Minimum Required Hardware

1x NVIDIA L4 24GB

9.0 / 21.6 GB usable

1x RTX 3060 12GB

9.0 / 10.8 GB usable

Apple M4 24GB

9.0 / 18.0 GB usable

Determine requirements for your specific workload:

Open in VRAM Calculator

Architecture Diagram

Noisy Video Latents3D Spatio-Temporal VAE Latents3D Patch Embedding3D Spatio-Temporal PatchifyHidden: 2k · Text: 6.36B enc · Timestep (t)x 54 DiT blocksAdaLN ModulationNorm · Timestep scale/shiftSpatio-Temporal Attention16 headsHead dim: 128+AdaLN ModulationNorm · Timestep scale/shiftFeed-Forward (MLP)Activation+Final Norm (AdaLN Scale)Predicted Velocity (v)Denoised 3D Latents to VAE

About HunyuanVideo 1.5 720P T2V

HunyuanVideo 1.5 is a 13B parameter open-weight video foundation model by Tencent, featuring a dual-stream DiT architecture, Qwen2.5-VL text encoder, and 3D VAE for cinematic realism.

Technical Specifications

Attention

Attention Heads

16

Attention Head Dimension

128

Dimensions

Auxiliary Parameters

6.36B

Hidden Dimension Size

2,048

Number of Layers

54

About HunyuanVideo


Other HunyuanVideo Models
  • No related models available