What Nous Research does
Nous Research is a frontier open-source AI lab building the Hermes family of open language models plus software and infrastructure aimed at βdistributed, unbiased trainingβ across the open internet. On its website, the company describes its core focus as language-model research (model architecture, data synthesis, fine-tuning, and reasoning) and training/infrastructure work to coordinate distributed runs.
More
Its most user-facing product is the Hermes Agent: an open-source agent that runs on a userβs machine (desktop) or other deployments, and that connects across multiple chat and communication surfaces. The Hermes Agent page describes built-in capabilities including persistent learning/memory, natural-language scheduling, delegating subtasks to isolated subagents, web browsing and automation, and multiple execution/sandbox backends (local, Docker, SSH, Singularity, Modal). To help developers and builders deploy and build around these models/agents, Nous also runs Nous Portal (a hosted product tiering and access layer for model/tool usage) and publishes research and releases on its website (including models, datasets, papers, and frameworks). Business model signals: Hermes Agent is offered under a free tier and paid tiers via Nous Portal; the Hermes Agent product page states that paid tiers include monthly credits and access to β300+β models, implying monetization through hosted access/credits rather than selling the base model weights. Strategically, Nous is positioning around two linked differentiation themes: (1) open-weight, steerable model releases (Hermes series) and (2) an infrastructure thesis for decentralizing training compute and coordination (Psyche/related tooling referenced in coverage and releases). In external reporting, the company is described as a decentralized open-source AI lab that leverages Solana primarily for coordination/incentives around distributed compute contributions.