Large vision-language models have demonstrated significant progress in complex reasoning tasks, but existing benchmarks primarily focus on single-image analysis, neglecting the potential of contextual information from multiple images. OMIBench is a newly introduced benchmark designed to assess the reasoning capabilities of these models in multimodal, multi-image contexts. By evaluating models on their ability to process and reason about multiple images, OMIBench provides a more comprehensive understanding of their capabilities. The development of OMIBench is crucial as it fills a gap in current benchmarking methods, allowing for a more accurate assessment of large vision-language models' abilities1. This advancement has significant implications for various fields, including policy, security, and workforce dynamics, as it highlights the potential applications and limitations of these models. The introduction of OMIBench enables researchers to better evaluate and improve the performance of large vision-language models, ultimately contributing to more effective and reliable AI systems.