Sand.ai Open-Sources 100B-Param Video Model MAGI-2 at 1/10th Inference Cost
量子位 · wechat · 2026-08-05
Tsinghua-backed team Sand.ai has released and open-sourced MAGI-2-preview, a 100-billion-parameter video generation model. Despite having 114B total parameters, it activates only about 6B parameters during a single forward pass through an extreme MoE architecture. Generating a 10-second 1080P video on 8x H100s costs only 0.5 RMB, roughly one-tenth the cost of mainstream industry models.
To overcome the communication and memory bottlenecks caused by ultra-long video sequences, the team built a custom infrastructure: a single-stream unified architecture for audio-visual processing, alongside a Multi-Head Latent MoE and Head Parallel strategy that decouples communication volume from the number of activated experts. The model ranks 6th on authoritative video generation leaderboards, closely trailing top-tier closed-source models.
More from Multimodal
- Pixio Multimodal Studio Integrates Image, Video, 3D and Music Generation — tsi_org · 2026-08-05
- ComfyUI Workflow: Video Lipsync Using Minimax and Exact Audio Lock Node — redkinoko · 2026-08-05
- Train AI Directly in Your DAW: M4LRhythmVAE Revamped as neu.rhythm — teropa · 2026-08-05
- Next-Gen AI Video: Minimax M3 Generates Surreal Vid2Vid Transformations — Promptmethus · 2026-08-05
- Test Drive: Qwen 3.8 Max Generates a Functional 3D Racing Game — iamfakhrealam · 2026-08-05
- ByteDance Launches SeedRealtime: Free Real-Time Video Chat in Doubao App — xiaohu · 2026-08-05