Atlas

Models

← All models

DeepHermes 3 Llama 3.1 8B Preview

Nous Research · 2025-02-12 · 8B parameters

DeepHermes 3 Llama 3.1 8B Preview is Nous Research's preview fine-tune of Meta's Llama 3.1 8B, released in February 2025. It is one of the first models to unify both standard conversational responses and deep chain-of-thought reasoning in a single model, with the reasoning mode toggled via system prompt.

Benchmark scores

BenchmarkScore
Artificial Analysis Intelligence Index2.3
Artificial Analysis Openness Index33.3
GPQA Diamond27.0
Humanity's Last Exam (Text-Only)4.3
LiveCodeBench (Artificial Analysis)8.5
MATH-500 (Artificial Analysis source)21.8
MMLU Pro36.5
SciCode (Artificial Analysis)9.1
Loading Atlas data…