# Show HN: We compressed a 31B video\+audio diffusion model to 14B, 6 steps, INT8

- HackerByte page: <https://hackerbyte.net/en/stories/compression-of-a-large-video-and-audio-diffusion-model-to-14b-parameters-using-i-50405263>
- Original source: <https://github.com/Video-Rebirth/hyperflow-h3-flash>
- Hacker News discussion: <https://news.ycombinator.com/item?id=50405263>
- Submitted to Hacker News: 2026-11-07T15:16:01.000Z

Compression of a large video and audio diffusion model to 14B parameters using INT8 quantization over 6 steps.

## AI-generated summary

No summary is available in this language yet.