TEST #2975 RESULTS
03/05/2026 - 10:49 AM
ACCELERATOR
MODEL
38.2
tokens/s
generation
1.52
sec
time to first token
951
tokens/s
prompt
288
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 9 5900X 12-Core Processor (znver3)
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
1133
tokens/s
45.0
tokens/s
926
ms
pp4096+tg256
716
tokens/s
22.2
tokens/s
5.76
sec
pp2048+tg256
952
tokens/s
33.5
tokens/s
2.18
sec
pp2048+tg768
950
tokens/s
31.4
tokens/s
2.18
sec
pp1024+tg1024
1122
tokens/s
38.2
tokens/s
936
ms
pp1280+tg3072
1077
tokens/s
28.2
tokens/s
1.21
sec
pp384+tg1152
1252
tokens/s
45.3
tokens/s
325
ms
pp64+tg1024
848
tokens/s
52.0
tokens/s
93
ms
pp16+tg1536
507
tokens/s
48.2
tokens/s
48
ms