Open-weight models built by the community. Hermes, DeepHermes, and the models powering private, steerable AI.
Nous Research is the community lab behind the Hermes family — open-weight models with strong instruction-following, tool use, and structured output. DeepHermes adds explicit reasoning. Everything is auditable and self-hostable.
Who it's for: Builders and researchers who want open, auditable weights they can self-host or run privately — not a black-box API.
The Hermes 2/3 family pairs strong general ability with tool use and structured output — fully open weights.
A reasoning variant that thinks before answering, rivaling closed models on logic benchmarks.
Run weights locally or on your own VPC — no data leaves your stack.
A simple API if you would rather not self-host, with the same open models behind it.
Nous Research is a pillar of the open-weight movement. If you care about owning your models and private inference, Hermes and DeepHermes are top-tier. Use the API for convenience or self-host for full control — either way, you are not locked in.