Peking University, Tsinghua and Alibaba Open-Source SparkDiffusion for Up to 265x Faster Wan Video Generation, Pandaily reports
Reported by Pandaily · not yet confirmed by the company or a second independent outlet. We update this page when it is.
Peking University, Tsinghua and Alibaba open-sourced SparkDiffusion, which cuts a 720P Wan2.1 14B video from 4,769 to 18 seconds on one RTX 5090, a 265x speedup. Code and weights are open, Pandaily reported.
China context
- Original name
- 阿里巴巴
- Outside China
- Open weights · pandaily.com
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Builders outside China can integrate SparkDiffusion to accelerate Wan video generation on consumer GPUs, reducing infrastructure costs.
- For investors
- Investors should monitor whether SparkDiffusion's efficiency gains lead to broader adoption of Wan-based video applications and increased competition in the open-weights video space.
SparkDiffusion pairs 97% sparse attention, 3-step distillation and FP8 to accelerate Wan video generation. The open-source release includes code and weights.
The 265x speedup is achieved through 97% sparse attention, 3-step distillation, and FP8 quantization, reducing generation time from 4,769 seconds to 18 seconds on a single RTX 5090 for 720P Wan2.1 14B.
Developers outside China can now run Wan video generation on a single consumer GPU in seconds instead of hours, lowering the hardware barrier for high-resolution video creation.
The open-source release may reduce inference costs for video generation, making it feasible for individual developers and small teams without large GPU clusters.
Next signals to check: adoption of SparkDiffusion in open-source video pipelines and whether the sparse attention and distillation techniques are applied to other models.