LinkS2Bench - Spatiotemporal Ordering: leaderboard

LinkS2Bench (LinkS²Bench, 2026) task on drone videos paired with satellite imagery: multiple-choice accuracy (%) on the order in which marked satellite regions appear in the video; higher is better.

Metric: Spatiotemporal Ordering: multiple-choice accuracy (%) on the order in which marked satellite regions appear in the video on the full LinkS2Bench (17,903 questions over 1,500 UAV videos paired with satellite images), temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Around October 2028. 18 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1LLaVA-OneVision-1.5-4B52.6#1855
2Gemini 3.1 Pro (Preview)44#99
3Seed 1.842.7#252
4GLM-4.6V-Flash42.1#1040
5Qwen 3.5 35B A3B34.7#492
6Llama 4 Scout31.3#1818
7Qwen 3.5 397B A17B26.1#264
8Kimi K2.525#266
9LLaVA-OneVision-1.5-8B23.7#1864
10GPT-5.4 Mini21.7#406
11Qwen 3.5 9B17.3#828
12GPT-5.416.4#135
13Ministral 3 3B14.9#3504
14Claude Sonnet 4.613.7#173
15Ministral 3 14B12.5#1870

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

Interactive version: theaggregate.ai/benchmark?slug=links2bench-spatiotemporal-ordering · How It Works · Data refreshed daily, snapshot 2026-11-21.