Tools › AI Models & ML › NVIDIA Cosmos

NVIDIA Cosmos

Open world models and datasets for robots and physical AI

🧪 AI Models & ML ★ 12.4k +8 this week Advanced

What it is

An open platform of world models, datasets and tools for physical AI, including robots, autonomous vehicles and smart infrastructure. It is aimed at robotics and physical-AI researchers who need generative simulation and action-aware models.

How it works

Cosmos provides an omni model that handles multiple input types, including images, text and spatial movement, and can predict future frames. It can be post-trained into a world action model that perceives, reasons, plans and generates actions for physical systems.

Use cases

  • Generate synthetic future frames for robot training and simulation.
  • Post-train a world action model for autonomous vehicle planning.
  • Experiment with multi-modal physical-AI datasets and tools.
  • Research perception and planning for smart infrastructure.

💰 Business opportunities

  • Consult on robotics simulation and synthetic data pipelines.
  • Build specialized physical-AI models for niche robotic tasks.
  • Offer integration services for companies exploring world models.

Featured in 1 video

Free clones: Notion, Lovable, Notebook LM + top 10 repos

The Next New Thing · 2026-06-12

Similar tools

MuJoCo

General-purpose physics simulator for robotics and control research

🧪 AI Models & ML ★ 18.3k +45 wk C++

Baidu Unlimited OCR

Long-document OCR with sliding-window attention

🧪 AI Models & ML

SkillSpector

Security scanner for third-party AI skills

🔒 Security & Privacy ★ 26.7k +156 wk Python

auto-ontology

A trusted semantic context layer so data agents stop guessing

📊 Data & Databases ★ 13.7k +253 wk Python

NVIDIA SwitchYard

Route LLM traffic across models to cut cost and keep quality

🧠 LLM Apps & Frameworks ★ 3.6k +25 wk Rust

Silo

Simulate a world to test AI agents before they go live

🤖 AI Agents & Assistants