Ross ROSS = Recommend OSS · open-source software intelligence for agents

NVIDIA/cosmos

NVIDIA Cosmos is an open platform of world models, datasets, and tools that enables developers to build Physical AI for robots, autonomous vehicles, smart infrastructure, and more. observed · 2026-08-28

github.com/NVIDIA/cosmos · homepage · Jupyter Notebook · NOASSERTION (other) observed · 2026-08-28

Health v2 · maintenance only

72/100

  • Activity 99
  • Release rhythm 54
  • Longevity 43

Flags: no_license

How is this computed?

round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10) — computed 2026-09-02. Adoption (stars, forks) is never an input.

  • gap_med: n/a
  • age_days: 611
  • days_rel: 93
  • days_push: 8
  • n_releases_24m: 1

Full methodology

Adoption not part of the score

11641 stars · 845 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

NVIDIA Cosmos is an open platform of omnimodal world foundation models, datasets, and tools for building Physical AI systems such as robots, autonomous vehicles, and smart infrastructure. Cosmos 3 unifies language, image, video, audio, and action generation in a Mixture-of-Transformers architecture, with integrations for Diffusers, vLLM, TensorRT-LLM, SGLang, and NIM.

Use cases

  • generate physics-grounded video for robot training
  • build world action models for robot policy learning
  • reason over camera feeds for traffic monitoring and quality inspection
  • simulate worlds to evaluate autonomous vehicle behavior
  • finetune a world foundation model on custom embodiment data
  • distill and export model checkpoints for deployment
  • run vision-language reasoning on real-world scenarios

When to choose

  • you need world foundation models for robotics or autonomous vehicle development
  • you want to post-train a generalist model on your own camera and embodiment data
  • you need controllable, physics-grounded video generation or world simulation
  • you want vision-language reasoning for real-time video analytics

When to avoid

  • you need a lightweight model that runs on CPU or consumer hardware without GPUs
  • you want a general-purpose chatbot rather than physical-world AI
  • your project requires a permissive license without NVIDIA's custom terms
  • you only need simple image classification or standard NLP tasks

Facets

framework · maturity active

machine-learning deep-learning llm-inference llm-training video-processing simulation data-science artificial-intelligence machine-learning deep-learning large-language-models robotics autonomous-vehicles simulation computer-vision python cloud world-models physical-ai foundation-models omnimodal robotics-simulation autonomous-driving diffusion-transformers nvidia mixture-of-transformers video-generation video gpu linux docker

10 sources

Member repositories

RepositoryRoleHealth v2
NVIDIA/cosmosmain72

For agents

markdown · JSON · MCP: product_card(name="NVIDIA/cosmos")

Data as of 2026-08-30T08:39:29.467469+00:00 · Report a problem