TEST #3963 RESULTS
07/29/2026 - 7:38 PM
ACCELERATOR
MODEL
46.3
tokens/s
generation
883
ms
time to first token
1465
tokens/s
prompt
425
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 9 3900X 12-Core Processor (znver2)
RAM
63.9GB
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
1753
tokens/s
48.7
tokens/s
605
ms
pp4096+tg256
1395
tokens/s
42.2
tokens/s
2.96
sec
pp2048+tg256
1604
tokens/s
46.1
tokens/s
1.30
sec
pp2048+tg768
1603
tokens/s
45.5
tokens/s
1.30
sec
pp1024+tg1024
1745
tokens/s
46.8
tokens/s
610
ms
pp1280+tg3072
1691
tokens/s
44.9
tokens/s
779
ms
pp384+tg1152
1818
tokens/s
47.6
tokens/s
233
ms
pp64+tg1024
1372
tokens/s
47.7
tokens/s
67
ms
pp16+tg1536
209
tokens/s
47.6
tokens/s
97
ms