# SenseNova 6.8 Code2Video



SenseTime



SenseTime's open-source SenseNova 6.8 Code2Video is a 27B model, fine-tuned from Qwen3.8-27B, that writes code to build and render short videos. Its Skill turns an idea into a video you can keep refining with follow-up requests, or turns reference videos into editable code you can remix. Apache 2.0.


## Availability

Open weights under Apache 2.0 on Hugging Face. Fine-tuned from Qwen3.8-27B \(27B parameters, 512K context\). The Skill, prompts and generation traces are open on GitHub, and you can serve it with vLLM, SGLang or LightLLM.
- Released: Oct 28, 2026 22:09 · Beijing Time (UTC+8)
- Review updated: Oct 29, 2026 10:30 Thu · Beijing Time (UTC+8)

## Sources

- Source excerpt: https://x.com/SenseTime_AI/status/2115445835068748044
- Source excerpt: https://huggingface.co/sensenova/SenseNova-6.8-Code2Video

Canonical entry: https://www.ailyre.com/source/?id=sensetime-sensenova-6-8-code2video&lang=en