Metric-Bench - Distance: leaderboard

Metric: Mean relative accuracy (%; inter-object distance questions; 1,340 template questions over ScanNetV2 indoor test scenes, each giving the real metric values of reference objects and asking for a target object's metric in camera coordinates; zero-shot; mean relative accuracy over thresholds 0.50-0.95; GPT-5-mini and Gemini-2.5-flash on a random 400-question subset). Source: arxiv.org. Saturation forecast: Around 2029. 15 models tracked.

Top models

#ModelScore
1GPT-5 Mini22.73
2Qwen 3 VL 8B21.14
3Qwen 2.5 VL 32B Instruct20.45
4Gemini 2.5 Flash18.87
5Qwen 2.5 VL 7B Instruct16.71

Interactive version: theaggregate.ai/benchmark?slug=metric-bench-distance · How It Works · Data refreshed daily, snapshot 2026-09-26.