Llama 3 Instruct 70B
Creator
Meta AI (FAIR)
Released
2024-04-18
Intelligence
8.9
Artificial Analysis Index
Coding
6.8
Artificial Analysis Index
In $/1M
$0.65
input tokens
Out $/1M
$2.75
output tokens
Blended $/1M
$1.18
3:1 blended
Speed
44
tokens / sec
Benchmark breakdown
Independent evaluation scores, normalized to 0–100.
GPQA Diamond
38
Humanity’s Last Exam
4
MMLU-Pro
57
SciCode
19
LiveCodeBench
20
MATH-500
48
AIME 2025—
τ²-Bench (agentic)
0
Terminal-Bench Hard
1
IFBench
37