TEST #1731 RESULTS
11/04/2025 - 9:27 AM
ACCELERATOR
MODEL
27.4
tokens/s
generation
689
ms
time to first token
2671
tokens/s
prompt
474
LocalScore
HOW YOU STACK UP
Explore All ResultsQwen2.5 14B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper PRO 9995WX
RAM
498.8GB
OS
Linux
Kernel Release
6.14.0-34-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #34-Ubuntu SMP PREEMPT_DYNAMIC Wed Sep 17 09:21:29 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
3665
tokens/s
29.8
tokens/s
314
ms
pp4096+tg256
2633
tokens/s
19.7
tokens/s
1.60
sec
pp2048+tg256
3208
tokens/s
25.3
tokens/s
678
ms
pp2048+tg768
3209
tokens/s
25.0
tokens/s
679
ms
pp1024+tg1024
3630
tokens/s
28.0
tokens/s
318
ms
pp1280+tg3072
3200
tokens/s
23.6
tokens/s
435
ms
pp384+tg1152
3439
tokens/s
30.7
tokens/s
142
ms
pp64+tg1024
1046
tokens/s
33.1
tokens/s
88
ms
pp16+tg1536
8
tokens/s
31.7
tokens/s
1.95
sec