Hermes 4 - Llama-3.1 405B
Non-reasoning
Creator
Nous Research
Released
2025-08-27
Intelligence
17.6
Artificial Analysis Index
Coding
18.1
Artificial Analysis Index
In $/1M
$1.00
input tokens
Out $/1M
$3.00
output tokens
Blended $/1M
$1.50
3:1 blended
Speed
39
tokens / sec
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
54
Humanity’s Last Exam
4
MMLU-Pro
73
SciCode
35
LiveCodeBench
55
MATH-500—
AIME 2025
15
τ²-Bench (agentic)
27
Terminal-Bench Hard
10
IFBench
35