🤖Nous Research Drops Hermes 4 Open Model That Beats ChatGPT Without Guardrails
TL;DR
Nous Research released Hermes 4, an open-weights model the team says outperforms ChatGPT on several benchmarks while shipping without the content restrictions of commercial frontier models. Hermes 4 is the highest-scoring fully open model of 2026 so far.
Nous Research released Hermes 4, an open-weights model the team says outperforms ChatGPT on several benchmarks while shipping without the content restrictions of commercial frontier models. Hermes 4 is the highest-scoring fully open model of 2026 so far.
Key Points
Open-weights model from Nous Research
Outperforms ChatGPT on multiple benchmarks
Ships without content restrictions — full responsibility on the deployer
Why It Matters
Hermes 4 reopens the unrestricted-open-model debate at frontier capability. Regulators and enterprise risk teams now have to decide whether 'capability without guardrails' counts as dual-use.
Quick Facts
Frequently Asked Questions
Why does this matter?
Hermes 4 reopens the unrestricted-open-model debate at frontier capability. Regulators and enterprise risk teams now have to decide whether 'capability without guardrails' counts as dual-use.
What happened?
Nous Research released Hermes 4, an open-weights model the team says outperforms ChatGPT on several benchmarks while shipping without the content restrictions of commercial frontier models. Hermes 4 is the highest-scoring fully open model of 2026 so far.
Comments
Be the first to comment
Enjoyed this article?
Get it daily. 7am. Free. Reads in 5 minutes.