What Nous Research does
Nous Research is a frontier open-source AI lab focused on (1) training and publishing open language models and (2) building infrastructure for distributed training and agent systems. On its website, the company says it coordinates “distributed, unbiased training” and frames its mission around advancing human rights and freedoms via the creation and broad availability of open source language models.
Its flagship applied product is Hermes Agent, an open-source self-hostable “agent that grows with you.” The product page describes Hermes as supporting multi-platform “connect” (including desktop installs and integrations such as Telegram, Discord, Slack, WhatsApp, Signal, email, and CLI), “remember” (persistent memory), “schedule” (focused automation), and “delegate” (isolated subagents with their own conversations/terminals/scripts), with support for “search” and “experiment” modes including sandboxed execution across multiple backends (local, Docker, SSH, Singularity, Modal).
For paid customers, Nous also operates Nous Portal, described on the Hermes Agent page as a subscription layer that includes monthly credits for Hermes Agent usage, access to a large model catalog, and built-in tool use.
A distinctive strategic pillar is the company’s distributed training/compute network storyline: Nous markets research and systems intended to let GPUs across the internet collaborate on training. Its DisTrO and related work are positioned as communication- and network-efficient distributed optimizers suitable for low-bandwidth, heterogeneous internet links.