VibePrevisBench - Screen - All: leaderboard
VibePrevisBench - Screen - All. VibePrevisBench frozen paper Table1:242 Blender4D tasks,94script39photo23continuation86video,agent writes/runs Blender Python. Scene score is rule-based84,406hidden facts; Screen score Gemini4Argon VLM judge (photo recognition only). Failed runs score0. All is weighted native aggregate; overall mean of SceneAll andScreenAll. Harness and exact effort are observation settings, not checkpoints. README values are native one-decimal report; JSON provides two-decimal aggregate separately. Metric: Score (%). Higher is better.
Metric: Score (%). Source: raw.githubusercontent.com. Saturation forecast: Around February 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 85.36 | #6 |
| 2 | GPT-6.1 Sol | 83.1 | #10 |
| 3 | Gemini 4 Argon | 81.66 | #11 |
| 4 | Claude Opus 5.5 | 79.77 | #5 |
| 5 | DeepSeek V4.1 Flash | 77.86 | #59 |
| 6 | GPT-5.6 Sol | 77.51 | #30 |
| 7 | Kimi K3 | 73.45 | #54 |
| 8 | GLM-5.3 Flash | 69.55 | #122 |
| 9 | Gemini 3.8 Flash | 68.73 | #36 |
| 10 | Seed 2.1 Pro | 66.23 | #135 |
| 11 | Muse Spark 1.3 | 64.96 | #46 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=vibeprevisbench-screen-all · How It Works · Data refreshed daily, snapshot 2026-11-13.