All tasks / Models
SoulX-FlashHead 1.3B
Open-weights 1.3B real-time streaming talking-head model with Lite and Pro checkpoints.
- Soul AI Lab
- Open weights
- Released 12 Feb 2026
Where it ranks
- –Avatars & dubbingModels · Local · no result on this task’s benchmarks yet
Facts
- Vendor
- Soul AI Lab
- Family
- SoulX-FlashHead
- Released
- 12 Feb 2026
- Weights
- Open: download and run it yourself
- License
- Apache-2.0
- Parameters
- 1.3B
- GPU memory to run locally
- ~24 GB (README: Lite model runs 96 FPS on a single RTX 4090 (24 GB); Pro is real-time on two RTX 5090; no minimum stated)
- 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
No tracked tool lists this model yet; as open weights it runs in local runtimes.