SandAI Unveils MAGI-2 Preview: 114B Parameter Open-Source Video Model
zephyr_z9 · x · 2026-08-14
SandAI has officially released the preview and tech report for MAGI-2, an ultra-fine-grained Mixture-of-Experts (MoE) video generation model. It boasts a total of 114B parameters but only activates 6B per token.
Key Highlights:
- Cost-Effective: Claims to generate at just 1/10th the cost of mainstream models.
- Fully Open-Source: Commits to 100% open-source weights and a comprehensive tech report.
- High Performance: Leveraging an ultra-fine-grained MoE architecture to deliver top-tier results while drastically lowering inference compute requirements.
Related event: SandAI Open-Sources MAGI-2, a 114B-Parameter MoE Video Model(4 posts)→
More from Models
- Kimi K3 checkpoint impresses: strong held-out results point to solid pretraining — SonglinYang4 · 2026-10-16
- 200 chess games suggest in-context learning doesn't work yet, experimenter says — akbirthko · 2026-10-16
- OpenAI's internal model shows real math "research taste," experts say its proofs go beyond prior literature — cong_ml · 2026-10-16
- User catches Opus spinning up throwaway zz- debug scripts, echoing the METR OAI-HuggingFace incident — holdenmatt · 2026-10-16
- OpenAI drops 722 math papers tied to unreleased model, mathematicians demand receipts — thursdai_pod · 2026-10-16
- Fable 5/5.1 models show a striking drive toward record keeping, observers report — repligate · 2026-10-16