**SegPoint: Segment Any Point Cloud via Large Language Model** This paper introduces SegPoint, a unified framework that leverages a multi-modal Large Language Model (LLM) to perform various 3D point cloud segmentation tasks, including 3D instruction segmentation, 3D referring segmentation, 3D semantic segmentation, and 3D open-vocabulary semantic segmentation. The authors address the limitations of existing methods, which often rely on explicit instructions and lack the ability to infer implicit user intentions. To evaluate the model's performance, they introduce the *Instruct3D* benchmark, featuring 2,565 point cloud-instruction pairs, and demonstrate that SegPoint achieves competitive results on established benchmarks such as ScanRefer and ScanNet, while excelling on the *Instruct3D* dataset. The key contributions include the integration of a Geometric Enhancer Module and Geometric-guided Feature Propagation to enhance geometric information and improve feature quality, respectively. The experimental results show that SegPoint is the first model to address these diverse segmentation tasks within a single framework, achieving satisfactory performance.**SegPoint: Segment Any Point Cloud via Large Language Model** This paper introduces SegPoint, a unified framework that leverages a multi-modal Large Language Model (LLM) to perform various 3D point cloud segmentation tasks, including 3D instruction segmentation, 3D referring segmentation, 3D semantic segmentation, and 3D open-vocabulary semantic segmentation. The authors address the limitations of existing methods, which often rely on explicit instructions and lack the ability to infer implicit user intentions. To evaluate the model's performance, they introduce the *Instruct3D* benchmark, featuring 2,565 point cloud-instruction pairs, and demonstrate that SegPoint achieves competitive results on established benchmarks such as ScanRefer and ScanNet, while excelling on the *Instruct3D* dataset. The key contributions include the integration of a Geometric Enhancer Module and Geometric-guided Feature Propagation to enhance geometric information and improve feature quality, respectively. The experimental results show that SegPoint is the first model to address these diverse segmentation tasks within a single framework, achieving satisfactory performance.