LongShOTBench - Multimodal: leaderboard

LongShOTBench (arXiv:2512.16978) Multimodal (MM) category breakdown: weighted criterion-level rubric score (0-100%) on questions requiring joint visual, audio, and speech integration over long videos; higher is better.

Metric: Weighted Rubric Score (%). Source: longshot.cvmbzuai.com. Saturation forecast: Around September 2028. 102 models tracked. Updated 21 Nov 2026.

Top models

#ModelScoreOverall rank
1Qwen 3 Omni 30B A3B (Thinking)66.97
2Gemini 3.1 Pro (Preview)59.63#99
3Qwen 3 VL 32B (Thinking)49.72#573
4Vero-Qwen3I-8B49.27
5Qwen 3 VL 235B A22B (Thinking)48.56#459
6Voxtral-Small-24B-250748.54#1246
7Voxtral-Mini-3B-250747.23#3682
8Kimi K2.646.83#192
9Qwen 3 VL 235B A22B Instruct46.01#532
10Qwen 3 VL 30B A3B (Thinking)45.87#719
11Qwen 3 VL 8B (Thinking)45.34#910
12Qwen 3 VL 32B Instruct44.34#606
13Intern-S1-mini44.28#1178
14Gemma 3 27B (IT)43.56#1324
15Gemma 3 12B (IT)42.07#1887

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

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