Sign in
Pending benchmarksTracking since Oct 28, 2026

Is SenseNova 6.8 Code2Video state of the art?Pending benchmarks.Announced Oct 28, 2026; no independent results yet.

SenseTime's open model for code-driven video creation: rather than generating pixels, it writes executable animation code, renders it into a short video, reviews the frames and revises, over long agent runs in Claude Code. Fine-tuned from Qwen3.8-27B. Released with its production skill, prompt templates and example trajectories; Apache-2.0 weights on Hugging Face.

Announced facts

The details so far

Lab
SenseTime
Announced
Oct 28, 2026
API
Not yet
Open weights
Published on Hugging Face under Apache-2.0 (sensenova org: repo created Oct 24, release commit Oct 25, docs update Oct 27, announced Oct 28)
Parameters
27B (Qwen3.8-27B base; 27.36B in the safetensors)
Base model
Qwen3.8-27B (fine-tune)
Type
Vision-language model that writes and renders code to make short videos; it does not generate video pixels directly
Workflow
A general model expands the prompt; Code2Video then builds the film in Claude Code with SenseTime's code-video skill and can apply revision notes
Context
512K tokens (524,288) in the recommended serving setup
Released with
Production skill and prompt templates on GitHub (OpenSenseNova/SenseNova-6.8-Code2Video) and example production trajectories with 357 figures

Tags marked unsure are not stated on the lab's own pages.

Next steps

What would let us call it

Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.

  • Publicly announced with a source
  • Usable by the public (API or weights)
  • Listed on an independent board we track
  • Enough results for a rank in a category
  • Holds across two snapshots (crowns only)
Self-reported

What the lab claims

Self-reported. Shown for context; not used in any verdict.

Lab claim: sustains production runs of more than 10 hours and more than 2,000 turns.

Lab claim: preliminary results show good visual composition and motion choreography in motion design.

Receipts

Sources

LABSenseTime announcement on Xx.com/SenseTime_AI/status/2115445835068748044 · Oct 28, 2026REPOSenseNova-6.8-Code2Video on Hugging Facehuggingface.co/sensenova/SenseNova-6.8-Code2VideoREPOOpenSenseNova/SenseNova-6.8-Code2Video on GitHub (skill and prompts)github.com/OpenSenseNova/SenseNova-6.8-Code2Video
Editorial

Compare when it lands

Unscored: its base model, Qwen3.8 27B. A scored comparison appears once both are on a shared board.

Browse releasesRSSShareJSON

Tracking continues until an independent board lists it.