TEST #1510 RESULTS
09/11/2025 - 10:28 PM
ACCELERATOR
MODEL
88.2
tokens/s
generation
447
ms
time to first token
2891
tokens/s
prompt
829
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper PRO 3945WX 12-Cores (znver2)
RAM
125.6GB
OS
Linux
Kernel Release
6.12.46-1-lts
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC Tue, 09 Sep 2025 18:30:25 +0000
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
3401
tokens/s
92.4
tokens/s
312
ms
pp4096+tg256
2772
tokens/s
78.9
tokens/s
1.49
sec
pp2048+tg256
3140
tokens/s
86.8
tokens/s
664
ms
pp2048+tg768
3151
tokens/s
85.9
tokens/s
662
ms
pp1024+tg1024
3414
tokens/s
89.9
tokens/s
311
ms
pp1280+tg3072
3303
tokens/s
83.8
tokens/s
398
ms
pp384+tg1152
3323
tokens/s
91.1
tokens/s
126
ms
pp64+tg1024
2737
tokens/s
93.1
tokens/s
33
ms
pp16+tg1536
778
tokens/s
92.3
tokens/s
30
ms