Meta's Segment Anything Model 3 detects, segments, and tracks objects in images and videos using simple text prompts or exemplars. Open weights and Segment Anything Playground available now.
Official announcement
Google DeepMind adds native vertical video generation with reference images – optimized for Shorts creators.
Open-source release enables text-prompt object detection, segmentation, and tracking across images and video.
xAI introduces 10-second 720p video clips with improved audio and motion from text prompts.
Boston Dynamics Atlas, LG CLOiD, Figure, and others demonstrate production-ready capabilities and AI integration.
Google releases enhanced reasoning models with faster performance and lower pricing.
Platform enforces stricter policies against mass-produced misleading AI videos and deceptive trailers.
Realism from Veo, Grok Imagine, Sora, and physical AI robots reshape content and robotics.
Tension grows as models like Veo, Gemini, and Grok use online content for training.
Roundup: Humanoid robot advances at CES, Gemini 3.1 updates, Grok Imagine video, and more.
Cosmos, GR00T, and reasoning VLAs advance robot learning and real-world deployment.