TEST #1354 RESULTS
08/10/2025 - 9:31 AM
ACCELERATOR
MODEL
44.9
tokens/s
generation
231
ms
time to first token
5483
tokens/s
prompt
1021
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper PRO 7965WX 24-Cores (znver4)
RAM
251.2GB
OS
Linux
Kernel Release
6.8.0-71-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #71-Ubuntu SMP PREEMPT_DYNAMIC Tue Jul 22 16:52:38 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
7029
tokens/s
50.6
tokens/s
165
ms
pp4096+tg256
6106
tokens/s
36.9
tokens/s
697
ms
pp2048+tg256
6676
tokens/s
48.3
tokens/s
328
ms
pp2048+tg768
6801
tokens/s
42.4
tokens/s
323
ms
pp1024+tg1024
7192
tokens/s
45.3
tokens/s
164
ms
pp1280+tg3072
6610
tokens/s
40.4
tokens/s
216
ms
pp384+tg1152
6601
tokens/s
46.0
tokens/s
79
ms
pp64+tg1024
1787
tokens/s
47.2
tokens/s
56
ms
pp16+tg1536
542
tokens/s
46.6
tokens/s
50
ms