TEST #3331 RESULTS
04/16/2026 - 9:03 AM
ACCELERATOR
MODEL
84.0
tokens/s
generation
378
ms
time to first token
3407
tokens/s
prompt
911
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
4178
tokens/s
87.3
tokens/s
257
ms
pp4096+tg256
3198
tokens/s
76.7
tokens/s
1.30
sec
pp2048+tg256
3828
tokens/s
83.1
tokens/s
548
ms
pp2048+tg768
3940
tokens/s
83.0
tokens/s
532
ms
pp1024+tg1024
4186
tokens/s
86.1
tokens/s
256
ms
pp1280+tg3072
3990
tokens/s
81.9
tokens/s
334
ms
pp384+tg1152
4180
tokens/s
83.1
tokens/s
105
ms
pp64+tg1024
2614
tokens/s
87.8
tokens/s
37
ms
pp16+tg1536
547
tokens/s
86.7
tokens/s
40
ms