Explore the companies, models, and tools behind American AI.
An independent directory of U.S. AI organizations and U.S.-led open models, software, and research. Explore what they build, what is available, and where to find the original sources.
Independent project. Not a United States government website.
Directory previews
Editor’s selection (Sep 29, 2026): These entries were picked to show the range of what the catalog covers — a nonprofit research lab, a chip designer, a compute provider, a data platform, a robotics company, and an open-source foundation; and a model family, a framework, a runtime, a dataset, an evaluation tool, and a research stack. They are not ranked and are not the most important entries.
Directory
Companies & Labs
U.S. AI organizations, labs, research units, and foundations: what they build and how their products are delivered.
- Ai2 (Allen Institute for AI)Model developer · Research lab · Seattle, Washington
- NVIDIAChip designer · Model developer · Santa Clara, California
- CoreWeaveCloud provider · Compute infrastructure · Livingston, New Jersey
- DatabricksData platform · Enterprise software · San Francisco, California
- Physical IntelligenceModel developer · Research lab · San Francisco, California
- PyTorch FoundationOpen-source steward
Directory
Open Models & Tools
U.S.-led open models, software, datasets, and evaluation tools: what is public, under which license, and who maintains it.
- OLMoAi2 (Allen Institute for AI)Model family
- PyTorchPyTorch FoundationSoftware
- vLLMPyTorch Foundation (hosted project), vLLM core maintainers (project Technical Steering Committee)Software
- DolmaAi2 (Allen Institute for AI)Dataset
- LM Evaluation HarnessEleutherAIEvaluation tool
- openpiPhysical IntelligenceSoftware
Compare
Catalog coverage
These are counts of published records in this catalog — not measures of capability, and not a census of American AI. Family overviews are counted separately from the releases they summarize, and research units are counted separately from their parents.
Open the comparison workspace| Record type | Count | What is counted |
|---|---|---|
| Organization recordsEvery published organization record, including research units and subsidiaries listed separately. | 99 | Every published organization record, including research units and subsidiaries listed separately. |
| Top-level organizationsOrganization records without a parent organization in this catalog. | 91 | Organization records without a parent organization in this catalog. |
| Units and subsidiariesOrganization records that belong to a parent listed in this catalog. Not independent companies. | 8 | Organization records that belong to a parent listed in this catalog. Not independent companies. |
| Model familiesOverview records. Not added to release counts. | 43 | Overview records. Not added to release counts. |
| Model releasesSpecific releases assessed. | 77 | Specific releases assessed. |
| FrameworksSoftware libraries for building or training models. | 26 | Software libraries for building or training models. |
| RuntimesSoftware for serving or running models. | 8 | Software for serving or running models. |
| Research stacksReference training or research codebases. | 5 | Reference training or research codebases. |
| DatasetsPublished datasets. | 8 | Published datasets. |
| Evaluation toolsBenchmarks and evaluation harnesses. | 11 | Benchmarks and evaluation harnesses. |
Who maintains the open artifacts in this catalog
Organizations with the most open-artifact records here. This counts catalog records, not importance, quality, or total output.
- NVIDIA18
- Google (Alphabet)15
- Meta15
- OpenAI10
- Ai2 (Allen Institute for AI)9
- Hugging Face9
- Microsoft9
- The Linux Foundation7
Editorial log
Recently reviewed
Entries whose evidence an editor checked most recently. A review date is not a release date. See the changelog.
How this catalog works
Evidence first, independent by design
Sourced claims
Each statement links to the official page, model card, repository, or license that supports it, with the date it was checked. Unknown stays unknown.
Published eligibility rule
Entries qualify through documented U.S. headquarters, a U.S. nonprofit or lab, documented U.S. control, or U.S.-based project governance — applied the same way to everyone.
Openness by artifact, not by reputation
Models, software, datasets, and evaluation tools each get their own checklist. Model tiers follow USASI rubric v0.1, an editorial rubric, not a certification.
Independent project. Not a United States government website. USASI is not affiliated with any listed organization, and a listing is not an endorsement, a safety assessment, or a ranking. Read the methodology or the disclaimer.
Local corner · September 2026
People behind local AI
A small, rotating look at people whose public work helps others run AI models on their own hardware.
- Jeffrey MorganWorks on Ollama, an open-source tool for running open models on your own machine
- Joshua LochnerWorks on Transformers.js at Hugging Face, a library for running Transformers models in the browser
- Maxime LabonneHead of Post-Training at Liquid AI; author of the LLM Course and its quantization tutorials
- Omar SansevieroDeveloper Experience Lead at Google DeepMind, working on Gemma launches and developer guides
- Stella BidermanExecutive Director of EleutherAI; first author of the Pythia open model suite
Support Us
Help keep USASI useful.
Find the catalog useful? Leave an optional tip to support its upkeep. Tips never affect listings, coverage, or openness assessments.
Optional. No USASI account required. Payment takes place on the linked provider’s website (Buy Me a Coffee).
About supporting this project