NVIDIA Launches Cosmos 3 Edge - An Open Frontier Model for Local Deployment
NVIDIA has unveiled its new Cosmos 3 Edge model designed for local deployment on various devices. This model aids in training robots and autonomous vehicles, as well as in computer vision tasks.
NVIDIA Launches Cosmos 3 Edge
NVIDIA has announced the release of Cosmos 3 Edge - their open frontier model designed for local deployment on devices.
Cosmos 3 Edge helps robots learn and act, enables autonomous vehicles to understand road scenes and predict the intentions of traffic participants, and allows AI vision agents to reason based on real-time video streams for smart infrastructure tasks.
The model features 4 billion parameters and includes a 2 billion parameter reasoner based on Nemotron. It can be deployed on DGX Spark, NVIDIA Jetson, and other devices.
Cosmos 3 Edge combines autoregressive and diffusion transformer blocks through a shared multimodal attention mechanism, linking understanding, prediction, simulation, and action within a single model.
Among open models of comparable size, Cosmos 3 Edge ranks first in VANTAGE-Bench for visual data analysis.
As always, Cosmos 3 Edge is fully open. Model weights, post-training recipes, and code are available.
The model is already available on Hugging Face.
Why it matters
AnalysisThis model opens new possibilities for autonomous systems, enhancing their ability to learn and adapt in real-time. It also contributes to the advancement of computer vision and artificial intelligence technologies.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Nvidia's Monopoly and the CUDA Platform
An analysis of Nvidia's technological and economic dominance in the AI market. The CUDA (Compute Unified Device Architecture) platform, created in 2006, transformed ordinary gaming GPUs into the planet's primary computational tool, making it difficult for competitors like AMD and Intel to break this monopoly.
Four Resume Tips from OpenAI’s Head of Applied AI Development
Wulfie Bain, OpenAI’s head of applied AI development, shared four resume rules: keep it short, remove irrelevant details, highlight outcomes, and lead with the strongest achievements.
AI Radar: GPT-6.1 Terra, Gemini 4 Carbon, Mistral Large 4 and OpenAI/Anthropic pricing changes
A roundup of recent AI news: new LLM releases, OpenAI and Anthropic subscription changes, world models, video generation, and research updates.
JSON — Structured Data for Programs, APIs, and AI Agents
A complete hands-on guide to JSON: syntax specifications, serialization via parse and stringify, REST API consumption, JSON Schema validation, and structured outputs for LLMs.