Nous Hermes
Nous Research's open Hermes models — steerable, minimally filtered
Overview
Open-weight instruction models from Nous Research, the decentralised-AI lab Paradigm backed at a reported $1B valuation in 2025. Hermes 4 (Aug 2025, Llama 3.1 70B/405B) adds hybrid reasoning with some of the lowest refusal rates among open models.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 11 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
Use cases
ASEAN Perspective
Nous Hermes in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
The Hermes series is the closest thing open-source AI has to a house brand: Llama- and Mistral-based fine-tunes known for strong instruction-following, steerability and refusal rates far below the big labs' defaults. Hermes 4 (August 2025, 70B and 405B) added hybrid reasoning, and Nous is no longer just a Discord collective — Paradigm led a $50M Series A in April 2025 at a reported $1B token valuation, and there's now a first-party chat app, hosted API and the Forge agent platform. Caveats: the crypto financing and Solana-based Psyche training network won't suit every procurement team, the 405B flagship is impractical to self-host, and quality varies by release and base model. For teams that want capable, minimally filtered open weights under their own control, Hermes remains the default answer.
What people say
Paradigm led a $50 million Series A into Nous Research in April 2025, at a reported $1 billion token valuation — a strange sentence to write about an outfit that started as a pseudonymous Discord collective fine-tuning Llama checkpoints. But the Hermes line earned it. From OpenHermes 2.5 through Hermes 2 Pro (which brought reliable function calling and JSON mode to small open models) to Hermes 3 on Llama 3.1 in 2024, each release has been among the most-downloaded fine-tunes of its generation.
Hermes 4, released August 2025 in 70B and 405B variants, added hybrid reasoning — toggleable chain-of-thought — and doubled down on the series' signature: low refusal rates. Nous positions Hermes as a neutrally aligned model that follows your system prompt rather than a corporate policy layer, and that steerability is why the local-model crowd keeps coming back to it for agents, roleplay and research that mainstream assistants decline.
The caveats have shifted. The old knock — community project, no first-party anything — no longer holds: there's a Nous Chat app, a hosted API, and the Forge agent platform. The newer concerns are different. Financing runs through crypto (the Series A was a token deal, and Psyche, the distributed training network, runs on Solana), which some procurement teams will read as a red flag. The 405B flagship is unrealistic to self-host, so most usage flows through OpenRouter or Nous's own endpoints anyway. And quality still varies by release and base model — pick your Hermes carefully.
For minimally filtered open weights you can actually control, this is still the default recommendation.
Summary of public user & expert reviews, compiled by RECATOOLS.
Notable facts
- Nous Research is a fully decentralised research collective with no official company registration — entirely volunteer contributors working open source.
- Hermes 3 achieved performance comparable to Llama 3.1 8B Instruct while having fewer refusals on legitimate edge-case questions.
- The Nous Research community developed the synthetic training data technique of using GPT-4 to generate diverse instruction datasets, which became a standard approach across the open-source model fine-tuning community.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including Nous Hermes's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Nous Hermes unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Nous Hermes directly →
Spotted something out of date? Suggest an update →
Nous Hermes in the news
Alternatives to Nous Hermes
More in LLMs & Chat
