# Hermes Agent Hits $1.5B: What Open-Source AI Gets Next > Source: > Published: 2026-07-14 06:13:58+00:00 Nous Research is raising at least $75M at a $1.5B valuation for [Hermes Agent](https://github.com/NousResearch/hermes-agent), an open-source autonomous AI agent that has hit 214,000 GitHub stars in under six months. [TechCrunch reported the round on July 13](https://techcrunch.com/2026/07/13/hermes-agent-maker-nous-research-in-talks-for-new-funding-at-1-5b-valuation/), led by Robot Ventures with USV participation — less than five months after a $50M Series A in February. The valuation puts Hermes in the same tier as established AI infrastructure companies, which is remarkable for a product launched this year. ## What Hermes Agent Actually Does Most AI agents forget everything the moment a session ends. Hermes does not. It runs a closed learning loop: after completing complex tasks, it writes reusable “skill” files that encode what worked and why. Those skills self-improve during subsequent use, and a full-text search system gives the agent accurate cross-session recall without relying on a vector database. The longer you run it, the faster it gets at your specific workflows. That sounds like marketing, so here is a concrete signal: a 2026 Nous Research survey of over 2,000 developers found that 73% abandoned AI agent tools within their first month due to lack of persistent memory. Hermes was built specifically around that failure mode. Developer-reported results after 30 days show a 1.6x task completion speed increase — not because the underlying model got better, but because Hermes stopped forgetting. Version 0.18.0, shipped July 1 and dubbed “The Judgment Release,” cleared every P0 and P1 bug across the entire repo — a meaningful signal that the project has moved from experimental to production-ready. Related:[AgentPrizm: Persistent Agent Memory via MCP Is Here] ## Getting Started Takes One Command Barrier to entry matters for adoption, and Hermes Agent has almost none. Installation on Linux, macOS, or WSL2 runs as a single command: ``` curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash ``` From there, `hermes setup` configures your model backend — Nous Portal, OpenRouter, OpenAI, or a custom endpoint — with no code changes required to switch between them. The `hermes gateway` command wires up messaging platform delivery across Telegram, Discord, Slack, WhatsApp, and Signal from a single process. The agent runs on a $5/month VPS, a serverless Modal instance that costs near-zero when idle, or your local machine. It is not tied to your laptop, which means it keeps working while you are offline. [The official documentation](https://hermes-agent.nousresearch.com/docs/) covers all six deployment backends in detail. ## The Open-Source Question That $1.5B Creates Hermes is MIT-licensed and Nous Research calls it “free forever.” That is true right now. However, $125M raised across two rounds in five months at a $1.5B valuation does not get paid back by selling to solo developers on $5 VPS servers. The cloud-hosted Nous Portal tier, priced at $20-$200/month, is where the revenue lives, and VC math eventually pushes teams toward shipping features there first. The pattern is familiar. Elastic, HashiCorp, and MongoDB all launched as MIT-licensed and open source. All of them drew lines between what is free and what requires a paid tier. None became worse products — but the calculus for self-hosting teams changed over time. Hermes’ self-hosted build will likely stay functional, but whether it stays on feature parity with the cloud tier is the real thing to watch. [Together AI’s $800M raise last week](https://byteiota.com/together-ai-raises-800m-open-source-inference-just-got-serious/) raised the same question about open-source inference infrastructure. There is also a governance note worth flagging: in May 2026, Nous Research faced community allegations that Hermes borrowed heavily from EvoMap’s Evolver project without attribution. GitHub issues documenting the claims were edited to remove them, and critics were blocked. Nous Research provided no formal response. That is not a dealbreaker for trying the tool — but it is a relevant data point if you are evaluating Hermes for long-term infrastructure use. ## Key Takeaways - Hermes Agent is a self-hosted, open-source AI agent that builds persistent skills from experience — 214,000 GitHub stars in six months reflects real developer demand, not hype. - Version 0.18.0 cleared all critical and high-priority bugs; this is now production-ready, not experimental. - One-command install, model-agnostic backend, runs on a $5 VPS — the trial cost is near-zero if you have been burned by stateless agents before. - The MIT license is real for now, but $125M in VC backing will eventually pressure the team toward cloud-tier feature velocity — watch the changelog for signs of divergence. - The plagiarism controversy and absence of a formal response from Nous Research are governance flags — do your own due diligence before making Hermes core infrastructure.