TEST #3479 RESULTS
05/09/2026 - 1:31 PM
ACCELERATOR
MODEL
79.2
tokens/s
generation
395
ms
time to first token
3204
tokens/s
prompt
863
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
127.9GB
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
3882
tokens/s
80.4
tokens/s
277
ms
pp4096+tg256
3136
tokens/s
73.5
tokens/s
1.32
sec
pp2048+tg256
3597
tokens/s
77.9
tokens/s
581
ms
pp2048+tg768
3696
tokens/s
79.2
tokens/s
566
ms
pp1024+tg1024
3811
tokens/s
79.1
tokens/s
281
ms
pp1280+tg3072
3833
tokens/s
77.3
tokens/s
345
ms
pp384+tg1152
4091
tokens/s
80.8
tokens/s
106
ms
pp64+tg1024
2186
tokens/s
83.0
tokens/s
42
ms
pp16+tg1536
606
tokens/s
82.0
tokens/s
38
ms