OST-Bench - Agent Visible Info: leaderboard

The Agent Visible Info questions of OST-Bench's multi-round version: the model recalls which objects it has seen, how many and where in the exploration. 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-4.176.46#505
2InternVL2.5-78B74.02#1036
3InternVL2.5-38B73.88#1366
4Gemini 2.0 Flash70.82#723
5Qwen 2.5 VL 32B Instruct64.9#1074
6Qwen 2.5 VL 72B Instruct64.46#849
7LLaVA-Video-72B-Qwen260.48#1344
8InternVL2.5-8B58.6#2401
9LLaVA-Video-7B-Qwen258.32#1910
10Qwen 2.5 VL 7B Instruct52.56#1893

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-visible-info · How It Works · Data refreshed daily, snapshot 2026-11-17.