TEST #2956 RESULTS
03/02/2026 - 12:34 PM
ACCELERATOR
MODEL
80.3
tokens/s
generation
411
ms
time to first token
3123
tokens/s
prompt
848
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper PRO 5945WX 12-Cores (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
3759
tokens/s
82.6
tokens/s
285
ms
pp4096+tg256
3015
tokens/s
74.2
tokens/s
1.37
sec
pp2048+tg256
3467
tokens/s
79.9
tokens/s
604
ms
pp2048+tg768
3490
tokens/s
79.3
tokens/s
600
ms
pp1024+tg1024
3753
tokens/s
81.6
tokens/s
285
ms
pp1280+tg3072
3613
tokens/s
77.3
tokens/s
367
ms
pp384+tg1152
3778
tokens/s
81.9
tokens/s
114
ms
pp64+tg1024
2579
tokens/s
83.4
tokens/s
36
ms
pp16+tg1536
653
tokens/s
82.6
tokens/s
36
ms