FastWan video generation model achieves 70x speedup, generates video in 5 seconds
Original title:Try FastWan at https://fastwan.fastvideo.org/! https://t.co/3dijljxKuI
AI summary
The FastVideo team launched the FastWan series of fast video generation models. The model uses a new training method called sparse distillation, which can increase video denoising speed by 70 times. On a single H200 GPU, it takes only 5 seconds to generate a 5-second video. The team provided an online demo and fully open-sourced the model, code, and data under the Apache-2.0 license.
Key Takeaways
- The FastVideo team launched the FastWan series of fast video generation models.
- The model uses a new training method called sparse distillation,which can increase video denoising…
- On a single H200 GPU,it takes only 5 seconds to generate a 5-second video.
- The team provided an online demo and fully open-sourced the model,code,and data under the Apache-2.
- 0 license.
Quick Facts
- Published
- Updated
- Category
- Model Releases
- AI score
- 67
视频生成终于从「等一分钟」进化到「实时出片」,FastWan 用稀疏蒸馏把去噪压了 70 倍,单卡 H200 五秒出五秒视频,做短视频工具和实时交互产品的团队该认真看看这个开源方案。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: FastWan video generation model achieves 70x speedup, generates video in 5 seconds. August 5, 2025. https://new.nat6.net/en/news/cmnxjn85o00h2sl9obw8o7gox/