SYNCR benchmark shows leading MLLMs reach only 52.5% average accuracy on cross-video reasoning tasks against an 89.5% human baseline, with major weaknesses in physical and spatial reasoning.
most and second most have the same number of collisions – 0
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SYNCR: A Cross-Video Reasoning Benchmark with Synthetic Grounding
SYNCR benchmark shows leading MLLMs reach only 52.5% average accuracy on cross-video reasoning tasks against an 89.5% human baseline, with major weaknesses in physical and spatial reasoning.