MolmoPoint: Better Pointing for VLMs with Grounding Tokens
Open paper
Benchmark
Model
Rank
Results
computer-use-agents-on-screenspot-v2
MolmoPoint-GUI-8B (64crops)
#15
Accuracy: 93.9
computer-use-agents-on-screenspot-pro
MolmoPoint-GUI-8B
#17
Accuracy: 61.1
Rank counts only results with a code link.