RoboStressBench - State Understanding: leaderboard

Metric: Accuracy (%) on the 633 object-state-understanding multiple-choice questions, exact match on the predicted option, greedy decoding with at most 64 new tokens (API models: temperature 0, 64 output tokens), one unified instruction template; images of embodied scenes under physical visual stress (material, viewpoint, lighting, geometry); higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 16 models tracked.

Top models

#ModelScore
1Qwen 3.6 27B67.1
2Qwen 3.6 35B A3B67.1
3Qwen 3.5 27B63.3
4Qwen 3.5 35B A3B62.9
5Qwen 3.5 4B59.6
6Qwen 3.5 9B58.9
7Gemini 3.1 Pro (Preview)41.6
8InternVL3.5-8B41.5
9Molmo2-8B39.2
10GPT-5.538.9

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