Researchers propose MXAttention, a data-free post-training quantization framework for MXFP4 attention that introduces Universal Optimal Scaling (UOS) with a distribution-independent optimal scaling bo…
A new training-free framework called Surprise Forcing improves long-video generation by dynamically allocating resources to key visual evidence and skipping unnecessary denoising steps. Developed by r…
Researchers propose SIRUS, a training-free inference-time framework for concept-level unlearning in text-to-video (T2V) models, achieving 70.4% average forgetting success and 25.7% average frame hit o…
As of mid-2026, open-source AI video models like Wan 2.2, Open-Sora 2.0, and HunyuanVideo can run on consumer GPUs with 8GB to 24GB VRAM, generating 5-20 second clips at 720p or 1080p. However, leader…
Researchers introduced Sol Video Inference Engine, an agent-native full-stack acceleration framework for video diffusion models that achieves over 2x end-to-end speedup with near-lossless quality acro…
Researchers propose Steady-Forcing, a memory and training framework for autoregressive video diffusion models that balances spatial persistence and motion continuity in long-horizon nature video gener…