
Independent company · New York, United States
Nous Research
American open AI lab behind the Hermes models, the Hermes Agent harness and decentralized training work such as DisTrO and Psyche.
New York, United States
- Founded
- 2023
- Type
- Independent company
- Based in
- New York, United States
- Key papers
- 4
- Featured projects
- 7
About
Nous Research began in 2022 as a group of volunteers who met on Discord, GitHub and Twitter, and became a company in 2023. Reported co-founders include Jeffrey Quesnelle, Karan Malhotra, Teknium and Shivani Mitra, and Bowen Peng is also described as a co-founder.
The group first fine-tuned open models into the Hermes series and published YaRN, a context-extension method. It later released the DisTrO optimizer and the Psyche network for training over the public internet, and in February 2026 launched Hermes Agent, an open-source agent harness.
Funding includes a 50 million dollar Series A led by Paradigm in April 2025 and a 90 million dollar round announced in October 2026.
Mission and ambitionsNous states its mission as creating and spreading open access to intelligence, so that powerful AI is in the hands of the many.
Known for
- 01
Hermes open-weight models
- 02
Hermes Agent open-source harness
- 03
DisTrO and Psyche distributed training
- 04
YaRN context extension
Milestones
- 2024
Seed round
Nous closed a 5.2 million dollar seed round in January 2024.
- 2024
Hermes 3 and DisTrO
Hermes 3 on Llama 3.1 and the DisTrO optimizer were published in August 2024.
- 2025
Series A and Psyche
A 50 million dollar round led by Paradigm accompanied a test network for the Psyche training system in April 2025.
- 2025
Hermes 4
The Hermes 4 models followed in August 2025.
- 2026
Hermes Agent
An open-source agent released in February 2026 that reached about 214,000 GitHub stars by July.
Open source
Agent framework that grows with the user; the org's flagship repo.
Evolutionary self-improvement tooling for Hermes Agent.
Adapter for running Hermes as a managed agent.
Reference code and prompts for Hermes function calling.
Hermes 3 at 8B, the most downloaded Nous-authored model.
- hermes-function-calling-v1dataset
Function-calling and JSON-mode training data.
- json-mode-evaldataset
Small JSON-mode evaluation dataset.
Research papers
Notable releases
Openness
Open weights and code; model licences follow their base models.
People
- Jeffrey QuesnelleCo-founder
- Karan MalhotraCo-founder
- Bowen PengResearcher, described as co-founder
- Dillon RolnickChief executive as of October 2026
Listed from the lab's own pages and public profiles.
Contact and community
Contact details and links are on the lab's website (https://nousresearch.com); code questions go through issues on their GitHub organisation.
Public Discord linked from the lab's own website for community discussion and announcements.
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.



