← All models

Nous: Hermes 3 70B Instruct

Nous Research

open_source_balanced

Capabilities

Structured OutputLong Context

Strengths

  • Excellent long-context coherence (131K tokens)
  • Strong multi-turn conversation and roleplay
  • Structured output and agentic capabilities
  • Open weights enable full data privacy via self-hosting
  • Competitive pricing for 70B-class model

Weaknesses

  • Very slow inference speed (0.03 score)
  • Weak reasoning and analytical performance
  • Code generation significantly below tier
  • No vision/multimodal capabilities

Pricing

Input / 1M tokens

$0.7

Output / 1M tokens

$0.7

Context window

131.072k

Transparency

Open weights

10.0 / 10

Open training data

0.0 / 10

Open methodology

7.0 / 10

Licence openness

6.0 / 10

Provider disclosure

6.5 / 10

FMTI company score

N/A

Composite:5.5 / 10

Open-weight model based on LLaMA 3.1 70B. Nous Research publishes model weights and provides reasonable methodology documentation via LLaMA foundation. Training data composition not fully disclosed. Model can be self-hosted for full data privacy.

Sustainability

Inference energy

N/A

Training footprint

N/A

Provider infrastructure

5.0 / 10

Composite:5.0 / 10

70B parameter model has moderate inference costs. Nous Research sustainability practices not publicly documented. Self-hosting enables local inference, reducing reliance on cloud infrastructure.

Task Fitness

question answering
34%
code
29%
mathematics
51%
business communication
50%
vision
0%
analysis
27%
extraction
74%
long-form generation
72%
research
50%
reasoning
16%
summarisation
75%
translation
68%
conversation
78%