Nvidia Open-Sources Cosmos 3 in 32B and 8B, Launches Cosmos Coalition for Physical AI
Nvidia open-sourced Cosmos 3 at GTC Taipei, describing it as the world’s first fully open omnimodel for physical AI that combines native vision reasoning with world and action generation. The release includes Super and Nano variants at 32B and 8B, and Nvidia said it is providing model weights, code, datasets and post-training recipes.
Nvidia said Cosmos 3 leads open-model benchmarks spanning world generation, vision understanding and robot-action tasks. The company also introduced Isaac GR00T, which it described as its first open humanoid robot reference design for robotics research, and formed the Cosmos Coalition with Black Forest Labs, Runway, Skild AI, Agile Robots, LTX and Generalist to build and open-source frontier world models for physical AI.
From the sources (25 posts)
@nvidiaaiIntroducing Cosmos 3: Our latest frontier model for Physical AI Cosmos 3 is the world’s first fully open omnimodel with native vision reasoning, world and action generation. Today we’re releasing Super (32B) and Nano (8B) variants. https:
@nvidiaaiCosmos 3 ties everything together. Previous releases separated world generation, physical understanding, and controlled scene generation. Cosmos 3’s MoT architecture unifies these capabilities by pairing an autoregressive reasoner tower w
@nvidiaaiDelivering leading results on physical AI benchmarks among open models, it ranks first across @ArtificialAnlys, Physics-IQ, PAI-Bench and R-Bench for world generation accuracy, RoboLab and RoboArena for action policy and the VANTAGE-Bench a
@nvidiaai@ArtificialAnlys In addition to understanding and reasoning across modalities, Cosmos 3 excels at simulating physical environments, predicting future world states, and helping train robots to perform specific tasks. It can do subsecond vi
@theturingpostBreaking news: Cosmos 3 is here. They are attempting to do something completely new 🤯 Why is Physical AI much harder than building a chatbot? Understanding the world is not enough, robots need to predict it and act inside it. That's the
@nvidiaai@ArtificialAnlys Image-to-video generation is just as impressive. Input image: "Generate a 16:9 image from a dashcam view of a formula 1 racing event" Video prompt: "A high-speed racing event where a car navigates multiple winding turns"
@victormustarRT @NVIDIAAI: Introducing Cosmos 3: Our latest frontier model for Physical AI Cosmos 3 is the world’s first fully open omnimodel with nati…
@mervenoyannNVIDIA just dropped Cosmos 3 at GTC 🔥 closest thing to AGI as world model > it can reason, understand AND generate videos, images, actions, text > sota, comes in 16B, 65B, with datasets > diffusers support 🧨 > open license 🤗
@huggingfaceRT @NVIDIAAI: Introducing Cosmos 3: Our latest frontier model for Physical AI Cosmos 3 is the world’s first fully open omnimodel with nati…
@theturingpostRT @TheTuringPost: Breaking news: Cosmos 3 is here. They are attempting to do something completely new 🤯 Why is Physical AI much harder th…
@stocksavvyshay$NVDA launched Cosmos 3 calling it “the frontier of Physical AI” and positioning it as the foundation for robotics and autonomous systems. The lineup spans high-accuracy robotics training, fast video reasoning and real-time edge inference.
@victormustarRT @NVIDIAAI: @ArtificialAnlys As always, Cosmos 3 is fully open. This includes model weights and post-training recipes. Available now on…
@scaling01RT @NVIDIAAI: Introducing Cosmos 3: Our latest frontier model for Physical AI Cosmos 3 is the world’s first fully open omnimodel with nati…
@kimmonismus1/ NVIDIA just open-sourced Cosmos 3 at GTC Taipei! It's the first fully open "omnimodel" for physical AI - one model that understands the real world, predicts what happens next, and generates the actions a robot should take. Weights, cod
@kimmonismus2/ What changed: old Cosmos split the work across separate models — one to understand a scene, one to generate video, one for controlled simulation. Cosmos 3 fuses everything into a single Mixture-of-Transformers with two towers: → a reason
@huggingfaceRT @mli0603: This is THE moment of Physical AI! We are officially announcing Cosmos 3: Omnimodal World Models for Physical AI 🚀 - Cosmos…
@kimmonismus7/ NVIDIA also launched the Cosmos Coalition - with Black Forest Labs, Runway, Skild AI, Agile Robots, LTX and Generalist - to push open world models forward together. The bet: world models are becoming the intelligence layer for robots an
@kimmonismus8/ Jensen's framing: the "big bang of physical AI" is just around the corner. Hype aside, this is the most serious open-source move in robotics yet. Data was the wall. Cosmos 3 is NVIDIA trying to knock it down — in the open.
@mtsliveSITUATION DETECTED: Nvidia launched Cosmos 3, an open world foundation model for physical AI that combines vision reasoning, world simulation, and action generation in a single system.
@c_valenzuelabA lot more to come! We are excited to work closely with NVIDIA on the Cosmos Coalition
@theturingpostRT @TheTuringPost: Breaking news: Cosmos 3 is here. They are attempting to do something completely new 🤯 Why is Physical AI much harder th…
@tomlikesrobotsRT @runwayml: Introducing the Cosmos Coalition A new global initiative with NVIDIA and leading AI labs to build and open-source frontier w…
@yacinemtbRT @NVIDIAAI: Introducing Cosmos 3: Our latest frontier model for Physical AI Cosmos 3 is the world’s first fully open omnimodel with nati…
@stocksavvyshay$NVDA announced Isaac GR00T which is its first open humanoid robot reference design for robotics research. The nearly 6-foot humanoid runs on Unitree hardware, Sharpa hands, Jetson Thor compute and Nvidia Isaac software.
@brianroemmeleThis is very big. NVIDIA new open humanoid robot reference design built for robotics research. The NVIDIA Isaac GR00T Reference Humanoid Robot. The garage robot builders have a friend in NVIDA. Astute robot companies will be OPEN SOURCE