TEST #2550 RESULTS
02/06/2026 - 1:19 PM
ACCELERATOR
MODEL
51.4
tokens/s
generation
1.45
sec
time to first token
853
tokens/s
prompt
311
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
31.9GB
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
1044
tokens/s
60.4
tokens/s
997
ms
pp4096+tg256
854
tokens/s
35.3
tokens/s
4.83
sec
pp2048+tg256
969
tokens/s
43.0
tokens/s
2.14
sec
pp2048+tg768
971
tokens/s
41.1
tokens/s
2.13
sec
pp1024+tg1024
1012
tokens/s
50.8
tokens/s
1.03
sec
pp1280+tg3072
987
tokens/s
37.8
tokens/s
1.32
sec
pp384+tg1152
1016
tokens/s
60.3
tokens/s
393
ms
pp64+tg1024
345
tokens/s
69.7
tokens/s
198
ms
pp16+tg1536
477
tokens/s
64.6
tokens/s
45
ms