According to AI at Meta, SAM 3 is a newly launched unified AI model that enables detection, segmentation, and tracking of objects across both images and videos. This next-generation model introduces highly requested features such as text and exemplar prompts, allowing users to segment all objects of a specific target category efficiently. The integration of these functionalities supports a wider range of computer vision applications, making it easier for businesses to automate image and video analysis workflows. SAM 3 represents a significant advancement in multimodal AI, offering practical opportunities for industries like retail, security, and autonomous systems to improve object recognition and streamline visual data processing (Source: @AIatMeta on Twitter, 2025-11-19).

According to @AIatMeta, Meta has launched SAM 3, a unified AI model capable of object detection, segmentation, and tracking across both images and videos. SAM 3 introduces new features such as text and exemplar prompts, allowing users to segment all objects of a specified category efficiently. These enhancements address highly requested functionalities from the AI community. The learnings from SAM 3 will directly power new features in Meta AI and IG Edits apps, empowering creators with advanced segmentation tools and expanding business opportunities for AI-driven content creation and automation. Source: @AIatMeta (https://go.meta.me/591040)