MMSI-Bench: A Benchmark for Multi-Image Spatial Intelligence 事件

MMSI-Bench: A Benchmark for Multi-Image Spatial Intelligence arXiv:2505.23764v3 Announce Type: replace Abstract: Spatial intelligence is essential for multimodal large language models (MLLMs) operating in the complex physical world. Existing benchmarks, however, probe only single-image relations and thus fail to assess the multi-image spatial reasoning that real-world deployments demand. We introduce MMSI-Bench, a VQA benchmark dedicated to multi-image spatial intelligence. Six 3D-vision resear