TEST #3581 RESULTS
05/21/2026 - 1:02 AM
ACCELERATOR
MODEL
29.4
tokens/s
generation
1.58
sec
time to first token
832
tokens/s
prompt
249
LocalScore
HOW YOU STACK UP
Explore All ResultsQwen2.5 14B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2678 v3 @ 2.50GHz (haswell)
RAM
15.5GB
OS
Linux
Kernel Release
6.8.0-47-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #47~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Oct 2 16:16:55 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
963
tokens/s
30.7
tokens/s
1.10
sec
pp4096+tg256
788
tokens/s
26.0
tokens/s
5.23
sec
pp2048+tg256
887
tokens/s
28.8
tokens/s
2.34
sec
pp2048+tg768
885
tokens/s
28.4
tokens/s
2.35
sec
pp1024+tg1024
945
tokens/s
29.8
tokens/s
1.12
sec
pp1280+tg3072
920
tokens/s
27.7
tokens/s
1.42
sec
pp384+tg1152
986
tokens/s
30.7
tokens/s
421
ms
pp64+tg1024
828
tokens/s
31.4
tokens/s
123
ms
pp16+tg1536
286
tokens/s
31.0
tokens/s
87
ms