MOVA-720p

A foundation model designed for Image-to-Video-Audio (IT2VA) and Text-to-Video-Audio (T2VA) tasks, enabling simultaneous generation of high-fidelity video and synchronized audio. It addresses limitations of cascaded pipelines and proprietary systems by providing a fully open-source solution.  

Key Features:  

• Native Bimodal Generation: Generates video and audio in a single inference pass, ensuring precise synchronization (e.g., lip-sync and environment-aware sound effects).  
• Accurate lip synchronization and sound effects: The model demonstrates state-of-the-art results in multilingual speech synchronization and context-aware sound effect generation.
• Mixture-of-Experts (MoE) Design: Total 32B parameters, with 18B active during inference for efficient deployment.  
• Open-Source Framework: Releases model weights, inference code, training pipelines, and LoRA fine-tuning scripts.  
• Resolution: 720p

---

The model is a component of the video generation pipeline, consisting of:

• Text encoder: ~5.7B parameters,
• audio VAE: ~372M parameters, 
• audio DiT: ~1.4B parameters, 
• video DiT: ~28.6B parameters, 
• video VAE: ~127M parameters, 
• dual-tower bridge: ~2.7B parameters,

Total: ~38.8B parameters

---

For local running, authors recommends using at least 24GB GPU to generate a 8-second video at360p resolution (with offloading).

Announce Date: 29.01.2026
Parameters: 32B
Developer: OpenMOSS
Diffusers Version: 0.36.0
License: Apache 2.0

Use our pre-built public endpoints for free to test inference and explore MOVA-720p capabilities. You can obtain an API access token on the token management page after registration and verification.

Rent your own physically dedicated instance with hourly or long-term monthly billing.

We recommend deploying private instances in the following scenarios:

• maximize endpoint performance,
• enable full context for long sequences,
• ensure top-tier security for data processing in an isolated, dedicated environment,
• use custom weights, such as fine-tuned models or LoRA adapters.

Contact our dedicated neural networks support team at nn@immers.cloud or send your request to the sales department at sale@immers.cloud.