OST-Bench - Agent State: leaderboard

The Agent State questions of OST-Bench's multi-round version: the model tracks its own position and orientation as it explores a scene frame by frame. Scores are accuracy (%); the README states 1.4k scenes and 10k question-answer pairs, the per-model sample is not stated.

Metric: Accuracy (%). Source: github.com. Saturation forecast: Estimated already saturated. 10 models tracked. Updated 17 Nov 2026.

Top models

#ModelScoreOverall rank
1GPT-4.147.23#505
2InternVL2.5-78B46.45#1036
3InternVL2.5-38B45.38#1366
4Gemini 2.0 Flash45.05#723
5Qwen 2.5 VL 32B Instruct43.75#1074
6Qwen 2.5 VL 72B Instruct43.48#849
7InternVL2.5-8B41.87#2401
8Qwen 2.5 VL 7B Instruct40.43#1893
9LLaVA-Video-72B-Qwen239.95#1344
10LLaVA-Video-7B-Qwen233.5#1910

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=ost-bench-agent-state · How It Works · Data refreshed daily, snapshot 2026-11-17.