SYNCR - Spatial Measurement: leaderboard

Accuracy (%) on measuring positions and distances across views in SYNCR (2026), a synthetic benchmark of multi-video questions with programmatically verified answers; zero-shot four-option multiple choice, exact match, 200 questions per task (chance 25). Higher is better.

Metric: Exact-match accuracy (%) on the SYNCR Spatial Measurement task (measuring positions and distances across views), zero-shot four-option multiple choice over 2 to 4 labelled input videos, exact match, deterministic decoding, on the paper's balanced subset of 200 questions per task; chance is 25; higher is better. Source: arxiv.org. Saturation forecast: Around 2031. 18 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1InternVL3.5-38B52.5#1026
2Gemini 3 Flash48#180
3InternVL3.5-2B46#3313
4InternVL3.5-14B45.5#1519
5Qwen 3 VL 32B Instruct45#607
6GPT-5.444#138
7Qwen 2.5 VL 7B Instruct44#1919
8Qwen 2.5 VL 72B Instruct44#849
9Qwen 3 VL 4B Instruct43.5#1348
10Qwen 3 VL 8B Instruct41.5#1043
11LLaVA-OV-72B41.5#1894
12Qwen 2.5 VL 32B Instruct40#1065
13InternVL3.5-1B39.5#6140
14InternVL3.5-4B39#1946
15InternVL3.5-8B37.5#1618

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=syncr-spatial-measurement · How It Works · Data refreshed daily, snapshot 2026-11-22.