Model Terminal

Nous: Hermes 3 405B Instruct

Hermes 3 405B Instruct is a full-parameter instruction fine-tune of Meta's Llama 3.1 405B base model, developed by Nous Research to deliver stronger user alignment, agentic reasoning, multi-turn conversation, tool use, and creative/roleplay capabilities. It is one of the largest open-weight instruct models released by an independent AI research lab. Source: supabase.

Value score
Not scored
Context
128K
tokens
Max output
tokens
Price

Capability radar

Identity

Developer
Nous Research
Openness
Open Weights
Modalities
Release
2024-07-23
Knowledge cutoff
Deprecation

Benchmarks

  • MMLU88.6 · 2024-07-23 · source

Compare nearby

Pricing

Input / 1M
Output / 1M
Speed