Agnes-AI / Agnes-Video-2.0-MoE

Other08Updated Oct 20, 2026

Downloads

0

Likes

8

Last updated

Oct 20, 2026

License

apache-2.0

Agnes-Video-2.0-MoE is an audio-video generation model based on DaVinci-MagiHuman. It accepts text, video, and audio inputs and generates video and audio. With TIRx kernels, a 5-second 480×272 clip at 32 denoising steps now takes 13.641 s on an NVIDIA B200.

Tags

  • audio-video-generation
  • multimodal
  • sparse-moe
  • video-generation

Technical specifications

ParametersNot available
LibraryNot available
LanguagesNot available
Licenseapache-2.0