LongShOTBench - Long Videos: leaderboard

LongShOTBench (arXiv:2512.16978) Long Videos duration slice: weighted criterion-level rubric score (0-100%) on long-duration omni-modal videos; higher is better.

Metric: Weighted Rubric Score (%). Source: longshot.cvmbzuai.com. Saturation forecast: Around 2029. 102 models tracked. Updated 21 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Qwen 3 Omni 30B A3B (Thinking)60.81
2Gemini 3.1 Pro (Preview)53.75#99
3Voxtral-Mini-3B-250748.89#3682
4Qwen 3 VL 235B A22B (Thinking)48.67#459
5Voxtral-Small-24B-250747.17#1246
6Qwen 3 VL 32B (Thinking)46.62#573
7Vero-Qwen3I-8B44.97
8Qwen 3 VL 30B A3B (Thinking)43.35#719
9Gemma 3 27B (IT)42.92#1324
10Kimi K2.642.51#192
11Qwen 3 VL 235B A22B Instruct42.21#532
12Intern-S1-mini41.31#1178
13Qwen 3 VL 8B (Thinking)40.97#910
14Gemma 3 12B (IT)40.38#1887
15Nemotron 3 Nano Omni 30B A3BReasoning39.65#1412

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=longshotbench-long-videos · How It Works · Data refreshed daily, snapshot 2026-11-21.