Hermes 4
Powerful reasoning model that's aligned to you
20 followers
Powerful reasoning model that's aligned to you
20 followers
Hermes 4 by Nous Research is a new steerable, open model based on Llama-3.1-405B. It's a hybrid reasoner that reduces refusals and aligns to the user, offering a more creative and less preachy AI experience.
Flowtica Scribe
Hi everyone!
Nous Research just released Hermes 4, a new open model that's aligned to you, not a rigid set of corporate rules.
It's a powerful hybrid reasoner that can think longer on hard problems, and it's posting some very strong results in math and coding.
I also have to say, the revamped Nous Chat UI is fantastic. That epic, orb-style interface is the kind of thing only Nous Research would build. A great release for those who value steerability and creative freedom.
This is amazing. Does Hermes 4 handle multi turn reasoning too?
The hybrid reasoner is intriguing how does it manage to balance creativity with factual accuracy?
Does Hermes 4 support multi-modal reasoning or is it currently limited to text? I’m curious if hybrid reasoner means it can handle a wider range of input types beyond just text prompts.
Does Hermes 4 come with function calling or tool use right out of the box?
How does Hermes 4 manage safety alignment compared to other open LLMs? While reducing refusals is fantastic for usability what measures are still in place to prevent misuse?
I love the direction you’re taking here. Do you think Hermes 4 will be more suited for personal projects and experimentation or do you see teams using it in production environments?