LongShOTBench - Information: leaderboard

LongShOTBench (arXiv:2512.16978) Information (Info) category breakdown: weighted criterion-level rubric score (0-100%) on factual information extraction and synthesis over long 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)66.15
2Gemini 3.1 Pro (Preview)57.58#99
3Voxtral-Mini-3B-250751.48#3682
4Voxtral-Small-24B-250749.93#1246
5Qwen 3 VL 235B A22B (Thinking)46.67#459
6Qwen 3 VL 32B (Thinking)45.73#573
7Kimi K2.643.78#192
8Vero-Qwen3I-8B43.75
9Nemotron 3 Nano Omni 30B A3BReasoning42.65#1412
10Qwen 3 VL 30B A3B (Thinking)42.38#719
11Qwen 3 VL 235B A22B Instruct41.22#532
12Gemma 3 27B (IT)40.54#1324
13Intern-S1-mini40.37#1178
14Qwen 3 VL 8B (Thinking)40.31#910
15Gemma 3 12B (IT)38.42#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-information · How It Works · Data refreshed daily, snapshot 2026-11-21.