Nous Research Hits $1.5B Valuation With $90M Series B to Launch Hermes Business AI Agents

TL;DR

• Nous Research has closed a $90 million Series B at a $1.5 billion valuation to take its open-source Hermes models from developer favorite to enterprise platform.
• The funding will scale decentralized training infrastructure, expand hiring, and launch Hermes Business AI Agents for sales, support, operations, and coding workflows.
• The move pits the crypto-native AI lab directly against OpenAI, Anthropic, and Cohere in the race for customizable, open-weight enterprise agents.

From Open-Source Darling to Enterprise Contender

Nous Research is no longer just the decentralized AI lab behind the cult-favorite Hermes models. The New York-based startup confirmed it has raised a $90 million Series B at a $1.5 billion valuation, marking a 3x jump from its $50 million Series A led by Paradigm in April 2025 and cementing it as one of the most valuable open-weight AI startups in the world.

The round signals a sharp pivot in strategy. Known for its research on decentralized training over Solana and its community-driven approach to building large language models, Nous is now channeling that momentum into the enterprise. Its next act is Hermes Business AI Agents, a suite of customizable agents aimed squarely at business users rather than just developers and researchers.

In a market dominated by closed giants, Nous is betting that open, tunable, and decentralized AI can win in the boardroom.

Where The $90 Million Will Go

According to the company, the Series B capital will be deployed across three core priorities: compute, talent, and go-to-market.

First is infrastructure for decentralized pretraining and fine-tuning. Nous pioneered methods for training large models across distributed, low-communication GPU clusters, including its Psyche network. The new funding will expand that network and secure additional centralized compute for its frontier Hermes models.

Second is hiring. The company plans to roughly double its 40-person team over the next 12 months, with a focus on applied AI engineers, enterprise solutions architects, and safety researchers. Unlike many labs centralizing in San Francisco, Nous plans to remain distributed-first with hubs in New York, San Francisco, and Europe.

Third is enterprise commercialization. Until now, Nous has monetized lightly via APIs and community support. The Series B will fund a dedicated enterprise sales and partnerships team, SOC 2 and compliance work, and integrations with major cloud and data platforms.

Inside Hermes Business AI Agents

The centerpiece of the announcement is Hermes Business AI Agents, Nous's first full product suite built for non-technical business users.

Built on the latest Hermes 4 open-weight models, the agents are designed to be deployed inside a company's own cloud or VPC, a key differentiator for regulated industries worried about data leakage to closed APIs.

Early details point to four launch offerings:

Hermes Sales and Support agents that can resolve tickets, qualify leads, and draft personalized outreach while pulling live data from Salesforce, Zendesk, HubSpot, and Slack.

Hermes Ops agents for finance, legal, and HR teams that can summarize contracts, reconcile invoices, automate onboarding, and generate reports with citations back to internal documents.

Hermes Dev agents that bring Hermes coding capabilities into GitHub and VS Code for code review, refactoring, and test generation.

And a no-code Hermes Studio that lets business teams build, test, and deploy custom agents with company-specific knowledge bases, guardrails, and permission controls.

Crucially, all agents can be fine-tuned on proprietary data without that data flowing back into public models, addressing one of the biggest blockers to enterprise open-source adoption.

Why Open Weights Could Disrupt Enterprise AI

Nous's timing is aggressive. Enterprise spending on AI agents is projected to surge through 2027, but CIOs remain split between the power of frontier closed models from OpenAI and Anthropic and the control, cost, and customization of open models from Meta, Mistral, and Cohere.

Nous is positioning itself as a third path: frontier-class performance with open-weight flexibility plus decentralized infrastructure that lowers training and inference costs.

That pitch resonates with two trends. One, growing demand for sovereign and on-prem AI as European and financial-services clients seek alternatives to U.S. closed APIs. Two, frustration with vendor lock-in and per-seat agent pricing. Nous says customers can self-host Hermes agents for predictable costs and deeply customize behavior in ways closed assistants don't allow.

What This Means For The AI Race

A $1.5 billion valuation puts Nous in direct competition with Cohere, Mistral AI, and Perplexity for the enterprise open AI crown, while challenging the dominance of OpenAI's ChatGPT Enterprise and Anthropic's Claude for Work.

It also validates investor appetite for decentralized AI. Backers are betting that distributed training, crypto incentives for compute contribution, and community fine-tuning can produce models competitive with labs spending billions on centralized clusters.

Challenges remain. Nous will need to prove its models can match GPT-5-class and Claude-class reliability on complex, multi-step business workflows, meet enterprise security audits, and scale support beyond its developer community. Incumbents are shipping new agent platforms almost weekly.

But with $140 million raised to date, a loyal open-source following with millions of Hermes downloads, and now a war chest for enterprise execution, Nous Research has made its intentions clear: it doesn't just want to build cool models, it wants to run how business gets done.

AndroGuider Team

Articles written by the AndroGuider team. We try to make them thorough and informational while being easy to read.