Hermes 4 - Llama-3.1 405B
Reasoning
Creator
Nous Research
Released
2025-08-27
Intelligence
18.6
Artificial Analysis Index
Coding
16.0
Artificial Analysis Index
In $/1M
$1.00
input tokens
Out $/1M
$3.00
output tokens
Blended $/1M
$1.50
3:1 blended
Speed
42
tokens / sec
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
73
Humanity’s Last Exam
10
MMLU-Pro
83
SciCode
25
LiveCodeBench
69
MATH-500—
AIME 2025
70
τ²-Bench (agentic)
22
Terminal-Bench Hard
11
IFBench
33