Qwen-3D: A Generalist 3D Vision-Language Model for Spatial Understanding

Benchmark Model Rank Results
3d-visual-grounding-on-scanrefer-valQwen-3D-7B#2Acc@0.25: 62.8
3d-visual-grounding-on-scanrefer-valQwen-3D-3B#5Acc@0.5: 55.3
question-answering-on-sqa3dQwen-3D-7B#4AnswerExactMatch (Question Answering): 59.6
visual-question-answering-vqa-on-realworldqaQwen-3D-3B#83Accuracy: 60.4