Nous Research has secured US$90 million in new funding to advance an open, user-controlled approach to artificial intelligence. The October 2026 round brought together NVIDIA, Microsoft's M12, Samsung, Robot Ventures, Union Square Ventures, Y Combinator, and Menlo Ventures, among others. The company described the new capital as modest by modern fundraising standards, while noting that its culture has consistently delivered more with fewer resources and less waste than larger rivals.
Building an Open Alternative
Founded in 2023, Nous Research has positioned itself against what it describes as an inexorable trend toward centralized AI systems and walled gardens. Its leadership argues that restricted access to powerful technology imposes broad costs on the public while the benefits accrue to only a few. The new capital will help build an open, competitive, and auditable alternative that the company considers safer because it is not controlled by a small number of gatekeepers.
Post-Training and User Alignment
Nous Research focused on post-training as the layer where a small team could have the greatest impact. It fine-tuned Meta's Llama models to prioritize user alignment, meaning models adhere to the worldviews of individual users rather than values imposed from the top down. The team also pioneered synthetic data, defined open source tool calling, and proved that smaller development units can build fine-tuned models that compete with much larger laboratories.
Longer Context Through YaRN
In its early years, the company published YaRN, a method that gave open models much longer context windows with substantially less additional cost. Longer context allows a model to work across more of a conversation, document, or codebase at once, which the team believed would unlock agentic interactions. Throughout 2025, Nous Research built an agent stack to advance that vision and support more capable autonomous workflows across longer tasks.
Hermes Agent Momentum
The company's central product is Hermes Agent, an open source system that improves with use and lets users choose where it runs and which models it employs. It remembers projects, writes reusable skills as it works, and carries knowledge across sessions. Released under an MIT license in February 2026, Hermes has been cloned more than 24 million times and drives an estimated 2.5% of global token usage, making it one of the most widely used open source AI agents.
Enterprise Expansion
With the new funding and partnerships, Nous Research plans to build and deliver Hermes for Businesses. The enterprise offering is designed to give companies control over their intelligence stacks, reduce spending by selecting the best models at the best prices for their goals, and support safe and private AI scaling. It also aims to help businesses build institutional knowledge from the bottom up by empowering the talent they already have.
Nous Research's central principle is that people and companies using an AI system should maintain control over it. The company continues to build in the open, with the goal of helping users own their intelligence rather than rely on closed platforms. The US$90 million funding round signals investor confidence in that approach as demand grows for transparent, customizable, and trustworthy AI systems across consumer and enterprise markets.