Llama 3 Instruct 8B
Creator
Meta AI (FAIR)
Released
2024-04-18
Intelligence
6.4
Artificial Analysis Index
Coding
4.0
Artificial Analysis Index
In $/1M
$0.04
input tokens
Out $/1M
$0.14
output tokens
Blended $/1M
$0.07
3:1 blended
Speed
83
tokens / sec
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
30
Humanity’s Last Exam
5
MMLU-Pro
41
SciCode
12
LiveCodeBench
10
MATH-500
50
AIME 2025—
τ²-Bench (agentic)
0
Terminal-Bench Hard
0
IFBench
25