TEST #1705 RESULTS
10/26/2025 - 7:47 PM
ACCELERATOR
MODEL
120
tokens/s
generation
176
ms
time to first token
7594
tokens/s
prompt
1727
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper PRO 3995WX 64-Cores (znver2)
RAM
251.5GB
OS
Linux
Kernel Release
6.17.5-061705-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #202510231631 SMP PREEMPT_DYNAMIC Thu Oct 23 19:57:43 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
9584
tokens/s
121
tokens/s
115
ms
pp4096+tg256
6671
tokens/s
110
tokens/s
623
ms
pp2048+tg256
8408
tokens/s
118
tokens/s
252
ms
pp2048+tg768
8430
tokens/s
117
tokens/s
251
ms
pp1024+tg1024
9735
tokens/s
120
tokens/s
114
ms
pp1280+tg3072
9517
tokens/s
116
tokens/s
143
ms
pp384+tg1152
9760
tokens/s
124
tokens/s
47
ms
pp64+tg1024
4833
tokens/s
127
tokens/s
21
ms
pp16+tg1536
1411
tokens/s
125
tokens/s
19
ms