
Large company · Santa Clara, United States
NVIDIA
A GPU maker that also releases open models, datasets and tooling, from Nemotron language models to Cosmos world models and speech ASR.
Santa Clara, United States
- Founded
- 1993
- Type
- Large company
- Based in
- Santa Clara, United States
- Key papers
- 4
- Featured projects
- 8
Update NVIDIA is a large public company, listed here for its extensive open model and dataset releases. Not all of its products or models are open, and licence terms differ by model.
About
NVIDIA was founded in 1993 by Jensen Huang, Chris Malachowsky and Curtis Priem and is headquartered in Santa Clara, California. It is primarily a hardware and software platform company, and its open-model work sits alongside that business.
Through its Hugging Face organization and GitHub, NVIDIA publishes model checkpoints, large synthetic and curated datasets, and training and inference software. Main families include Nemotron for language and reasoning, Cosmos for physical-AI world models, Isaac GR00T for humanoid robots, and Parakeet, Canary and Magpie for speech.
Research is published as technical reports on arXiv, often with long author lists, and many releases come with training recipes and data. NVIDIA says its Nemotron effort includes open weights, open datasets and published methods.
Mission and ambitionsNVIDIA presents its open releases as a way to let developers and organizations build, customize and deploy AI on its platform, including sovereign and domain-specific systems. It describes Nemotron as open technology covering models, data and recipes.
Known for
- 01
Nemotron open language models and datasets
- 02
Cosmos world foundation models for physical AI
- 03
GR00T open humanoid robot models
- 04
NeMo toolkit with Parakeet and Canary speech models
Milestones
- 2019
Megatron-LM
Open research code for training large transformer models at scale, published on NVIDIA's GitHub.
- 2024
Canary-1B
Multilingual speech recognition and translation model from the NeMo family, released under CC-BY-NC-4.0.
- 2024
Nemotron-4 340B
Family of 340B-parameter models with a technical report; released under the NVIDIA Open Model License.
- 2025
Cosmos platform
World foundation model platform for physical AI, described in a January 2025 technical report.
- 2025
GR00T N1
Open foundation model for generalist humanoid robots, announced in March 2025.
- 2025
Nemotron-H and Nemotron 3
Hybrid Mamba-Transformer models, followed by the Nemotron 3 family of Nano, Super and Ultra announced in December 2025.
Open source
Open-source toolkit for speech models (ASR, TTS, speech LLMs), Apache-2.0.
Research code for training transformer models at scale.
Open platform of world models, datasets and tools for physical AI.
Primary Nemotron repository, listed as Apache-2.0 on NVIDIA's model info page.
600M-parameter English speech-to-text model under CC-BY-4.0.
- nvidia/canary-1bmodel
Multilingual ASR and translation model for English, German, French and Spanish (CC-BY-NC-4.0).
- nvidia/AeroZoodataset
Recently updated dataset on the NVIDIA organization page.
Demo space for NVIDIA's Nemotron multilingual transcription models.
Research papers
Notable releases
Openness
Many NVIDIA models ship as open weights, with datasets and code on Hugging Face and GitHub, but licences vary: the NVIDIA Open Model License covers Nemotron-4 and the Nemotron 3 Nano and Super models, newer Nemotron 3 Ultra and 3.5 models are reported under OpenMDW-1.1, NeMo code is Apache-2.0, and speech models range from CC-BY-4.0 (Parakeet) to CC-BY-NC-4.0 (Canary-1B).
People
- Bryan CatanzaroVice President, Applied Deep Learning Research; leads the Nemotron effort
Author of NVIDIA's blog post presenting Nemotron as open technology.
- Jensen HuangCo-founder
Co-founded NVIDIA in 1993; Wikipedia's Nvidia article currently reports a change in his CEO role.
- Linxi FanProject lead, GR00T N1
Known as Jim Fan; named as project lead on the GR00T N1 paper.
- Yuke ZhuProject lead, GR00T N1
Named as project lead on the GR00T N1 paper.
- Sanja FidlerResearcher, Cosmos
Listed among the Cosmos platform paper authors.
Listed from the lab's own pages and public profiles.
Contact and community
No general contact address was verified; use NVIDIA's Hugging Face and GitHub pages for model and repository issues.
No official NVIDIA Discord for open models was verified.
Sources
Compiled from the lab's public pages, Hugging Face and GitHub. Last checked 2026-11-17. Something wrong or out of date? Tell us on Discord.



