RoboStressBench - Planning: leaderboard

Metric: Accuracy (%) on the 821 planning 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 March 2028. 16 models tracked.

Top models

#ModelScore
1Qwen 3.5 35B A3B92.9
2Qwen 3.6 35B A3B81.1
3Qwen 3.5 27B77
4Qwen 3.5 4B68.8
5Qwen 3.6 27B68.8
6Qwen 3.5 9B65.5
7GPT-5.557
8Gemini 3.1 Pro (Preview)56.3
9Molmo2-8B54.3
10InternVL3.5-8B50.3

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