TEST #1820 RESULTS
11/17/2025 - 6:52 AM
ACCELERATOR
MODEL
49.6
tokens/s
generation
845
ms
time to first token
1555
tokens/s
prompt
450
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 5 3600 6-Core Processor (znver2)
RAM
47.9GB
OS
Windows
Kernel Release
10.0
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64
RUNTIME
Name
llamafile
Version
0.9.3
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1823
tokens/s
50.8
tokens/s
582
ms
pp4096+tg256
1441
tokens/s
45.4
tokens/s
2.86
sec
pp2048+tg256
1672
tokens/s
49.6
tokens/s
1.25
sec
pp2048+tg768
1671
tokens/s
49.0
tokens/s
1.25
sec
pp1024+tg1024
1816
tokens/s
50.2
tokens/s
584
ms
pp1280+tg3072
1757
tokens/s
47.9
tokens/s
748
ms
pp384+tg1152
1894
tokens/s
51.0
tokens/s
222
ms
pp64+tg1024
1447
tokens/s
51.5
tokens/s
63
ms
pp16+tg1536
471
tokens/s
51.2
tokens/s
54
ms