SenseNova 6.8 Code2Video

Open 27B model that writes code to render short motion graphics and explainers.

Also known as Code2Video, SenseNova

What it does

SenseNova 6.8 Code2Video is an open model from OpenSenseNova for code-driven video creation. The GitHub README says it is fine-tuned from Qwen3.8-27B, supports a 512K context, and builds short videos by generating executable code. Its limitations section says the model makes videos that can be written as code, including motion graphics, explainers, stylized 2D and 3D scenes, and procedural audio, and that it does not generate photorealistic footage. The README says the 90-second launch film, and every video in it, was made by the model. The Hugging Face model card describes the same approach, lists English and Chinese, and says the weights are Apache 2.0. The Hub API lists 27,356,728,560 BF16 parameters. The README badge says 27B. config.json sets max_position_embeddings to 262,144, rope_type to yarn, factor to 2.0, and original_max_position_embeddings to 262,144. Serving examples set the context length to 524,288. The README's recommended Claude Code settings include runs of more than 10 hours and more than 2,000 turns, and the repository ships a code-video skill with eight craft guides. Neither page lists a hosted price. Neither page prints a publication date. The GitHub repository and the README commit are from October 24, 2026, and the Hugging Face repository was last modified on October 27, 2026.

Who is SenseNova 6.8 Code2Video for?

Creators and developers who want short motion graphics, explainers, or stylized scenes rendered from code.

Tags

• open-weight
• video
• code