According to DeepLearning.AI, Meta has launched a comprehensive open-weights image segmentation suite featuring SAM 3 for segmenting images and videos—including from text prompts—SAM 3D Objects for converting segmented items into 3D meshes or gaussians using point clouds, and SAM 3D Body for generating full 3D human figures. Meta’s internal tests indicate these models surpass most competitors in both segmentation accuracy and 3D reconstruction quality. All models are accessible online with downloadable weights under the Meta license, offering businesses and developers practical tools for advanced computer vision and AI-driven content creation. (Source: DeepLearning.AI, Dec 8, 2025)