Nous Research raised a $90 million Series B at a $1.5 billion valuation and launched Hermes for Businesses on 7 October, bringing the startup’s total funding to $158 million, TechCrunch reported.
Key points
- Robot Ventures led the round, with Nvidia and Samsung among the investors named in two reports.
- Hermes for Businesses is intended to let companies deploy customised agents for multi-step work on their own data.
- Nous estimates that the open-source Hermes agent accounts for roughly 2.5% of global AI token usage.
- The startup expects annualised revenue to exceed $100 million before the end of 2026, up from roughly $36 million by mid-September.
Robot Ventures leads the Series B
Robot Ventures led the financing, with Nvidia, Union Square Ventures, Menlo Ventures, Samsung and 1789 Capital also participating, TechCrunch reported. The round takes the three-year-old company’s cumulative funding to $158 million. Nous developed Hermes as an open-source agent before moving to a product aimed at corporate users.
A separate VentureBurn account named Nvidia, Microsoft’s M12, Samsung, Robot Ventures, Union Square Ventures and Y Combinator as investors in the Series B. Its list includes M12 and Y Combinator, while TechCrunch’s includes Menlo Ventures and 1789 Capital. Both accounts name Robot Ventures, Nvidia, Samsung and Union Square Ventures.
The financing is directed towards bringing Hermes into businesses rather than solely distributing it to developers. The capital will support Hermes for Businesses, a product for deploying customised agents that carry out workflows with multiple steps. The company’s open-source agent has been cloned more than 24 million times.
Hermes for Businesses offers local deployment
Companies can run models locally or in the cloud through Hermes for Businesses, VentureBurn reported. It said users can choose underlying models according to their pricing and operational requirements. The product is aimed at companies seeking to use their own data while retaining control of where the models run.
VentureBurn described coding, image understanding and web search as functions of Hermes for Businesses. The enterprise offering is intended to let companies put customised agents to work on multi-step processes while keeping their data private.
Nous estimates that Hermes drives roughly 2.5% of global AI token usage. That is the startup’s estimate of use across AI systems, rather than a reported measure of revenue from the business product. VentureBurn dated the agent’s launch to February and also reported more than 24 million clones.
Hermes MoA 2.0 combines rival models
Nous released Hermes Mixture of Agents 2.0 on 5 July as an update to its open-source Hermes Agent framework, Yellow reported. The system sends a request to several models and assembles their responses. Users can configure which models take part, while its default selection includes GPT, Claude and DeepSeek.
Benchmark results shared with that release rated Hermes MoA 2.0 above each of its component models on reasoning, coding and instruction-following tasks, according to Yellow. Yellow said the system had not been tested against the most recently released frontier models and that Nous had not published a formal academic paper alongside it.
Nous targets $100 million in annualised revenue
Nous was generating roughly $36 million in annualised revenue by mid-September 2026. Annualised revenue expresses sales at a yearly rate. The figure precedes the launch of Hermes for Businesses reported on 7 October.
Nous expects to pass $100 million in annualised revenue before the end of 2026.
