LinkS2Bench - Zone Counting: leaderboard
Metric: Zone Counting: mean relative accuracy (%, MRA over thresholds 0.50-0.95 with a zero-count rule) on the number of vehicles or pedestrians in a marked satellite region 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: Not forecast. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 9B | 74.8 |
| 2 | Kimi K2.5 | 73.5 |
| 3 | Claude Sonnet 4.6 | 67.3 |
| 4 | Ministral 3 14B | 64.1 |
| 5 | Gemini 3.1 Pro (Preview) | 63.7 |
| 6 | Qwen 3.5 4B | 63.7 |
| 7 | GPT-5.4 | 63.6 |
| 8 | Ministral 3 3B | 61.9 |
| 9 | Ministral 3 8B | 59.1 |
| 10 | GPT-5.4 Mini | 54.4 |
| 11 | Qwen 3.5 35B A3B | 42.6 |
| 12 | Seed 1.8 | 41.1 |
| 13 | Qwen 3.5 397B A17B | 36.1 |
| 14 | Llama 4 Scout | 22.5 |
Interactive version: theaggregate.ai/benchmark?slug=links2bench-zone-counting · How It Works · Data refreshed daily, snapshot 2026-10-07.