OPEN LINK ↗
// 83d agoMODEL RELEASE
OpenMOSS unveils MOVA for synced video-audio generation
OpenMOSS released MOVA, an open-source 32B MoE (18B active) model for synchronized video and audio generation with lip-synced speech and aligned sound effects. The release includes a technical report plus open code, weights, and inference workflows, positioning MOVA as a rare open alternative to closed audiovisual generators.
// ANALYSIS
This is a meaningful open-source milestone in multimodal generation, especially for teams that need controllable end-to-end audio-video outputs instead of black-box APIs.
- –Joint video-audio generation in one pass can reduce sync drift and pipeline complexity versus cascaded setups.
- –Open weights and training/inference code make MOVA useful for research replication and downstream customization.
- –Support for 360p and 720p checkpoints gives developers a practical quality-vs-compute path.
- –The project directly targets a gap where most top-tier synchronized AV systems remain closed.
// TAGS
movamultimodalvideo-genaudio-genopen-sourceopen-weightsresearch
DISCOVERED
83d ago
2026-03-05
PUBLISHED
83d ago
2026-03-05
RELEVANCE
9/ 10
AUTHOR
AI Search