GST-Bench: leaderboard

Metric: Average score (%; unweighted mean of the 12 subtask scores (mean accuracy over tolerance thresholds for direction and position, mean relative accuracy for distance, accuracy for multiple choice); zero-shot, greedy decoding, official prompt templates, over human-verified questions on 6,790 minutes of simulator-rendered egocentric exploration video). Source: arxiv.org. Saturation forecast: Around April 2028. 22 models tracked.

Top models

#ModelScore
1Gemini 3 Pro42.68
2Gemini 2.5 Pro40.95
3GPT-540.85
4Seed 1.834.04
5Qwen 3 VL 32B30.43
6GPT-4o30.33
7InternVL3.5-8B22.99

Interactive version: theaggregate.ai/benchmark?slug=gst-bench · How It Works · Data refreshed daily, snapshot 2026-09-29.