MMOOC - Uncertain Spatial and Physical Context (Open-Ended VQA): leaderboard

MMOOC tests whether multimodal models decline questions their image and context cannot answer, and explain why. This board covers open-ended visual questions where the spatial or physical situation cannot be settled from the image, and averages the share of correct refusals with a judge rating of the abstention reasoning.

Metric: Out-of-context score (%; mean of the refusal rate, the share of the Uncertain Spatial and Physical Context Open-Ended VQA questions, which the image and context cannot answer, that the model correctly declines, and refusal rationality, the mean 0-1 judge score of its abstention reasoning in steps of 0.25; judge scores averaged over GPT-5.6, Claude Opus 5 and DeepSeek-V4-Pro). Source: arxiv.org. 18 models tracked.

Top models

Interactive version: theaggregate.ai/benchmark?slug=mmooc-uncertain-spatial-and-physical-context-open-ended-vqa · How It Works · Data refreshed daily, snapshot 2026-11-13.