LongShOTBench - Audio: leaderboard

LongShOTBench (arXiv:2512.16978) Audio modality slice: weighted criterion-level rubric score (0-100%) on ambient-audio-grounded questions over long 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

#ModelScoreOverall rank
1Qwen 3 Omni 30B A3B (Thinking)57.22
2Gemini 3.1 Pro (Preview)48.06#99
3Qwen 3 VL 235B A22B (Thinking)44.9#459
4Qwen 3 VL 32B (Thinking)44.14#573
5Voxtral-Mini-3B-250743.75#3682
6Vero-Qwen3I-8B43.58
7Voxtral-Small-24B-250741.86#1246
8Kimi K2.641.34#192
9Qwen 3 VL 30B A3B (Thinking)40.38#719
10Qwen 3 VL 235B A22B Instruct39.35#532
11Qwen 3 VL 8B (Thinking)38.71#910
12Gemma 3 27B (IT)38.66#1324
13Intern-S1-mini38.44#1178
14Qwen 3 VL 32B Instruct36.9#606
15Gemma 3 12B (IT)36.45#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-audio · How It Works · Data refreshed daily, snapshot 2026-11-21.