All tasks / Models
HuMo-17B
Open-weights human-centric video model conditioned on text, reference images and audio for talking and singing people.
- ByteDance (Intelligent Creation Lab)
- Open weights
- Released 10 Sep 2025
Where it ranks
- –Video generationModels · Local · no result on this task’s benchmarks yet
- –Avatars & dubbingModels · Local · no result on this task’s benchmarks yet
Facts
- Vendor
- ByteDance (Intelligent Creation Lab)
- Family
- HuMo
- Released
- 10 Sep 2025
- Weights
- Open: download and run it yourself
- License
- Apache-2.0
- Parameters
- 17B
- GPU memory to run locally
- ~24 GB (README: 17B runs on an RTX 3090 (24 GB) via ComfyUI-WanVideoWrapper; 1.7B variant needs a 32 GB GPU with the official code)
- Access
- Free (open weights)
- Last verified
- 19 Oct 2026
Price from github.com, 19 Oct 2026.
Benchmark results
No public benchmark results recorded yet.
Where you can use it
- SComfyUIWorkflow platforms · Local