TEST #3899 RESULTS
07/14/2026 - 2:13 AM
ACCELERATOR
MODEL
48.3
tokens/s
generation
3.61
sec
time to first token
387
tokens/s
prompt
173
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Core Ultra 9 285HX (arrowlake-s)
RAM
127.4GB
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
460
tokens/s
50.0
tokens/s
2.25
sec
pp4096+tg256
332
tokens/s
42.6
tokens/s
12.36
sec
pp2048+tg256
379
tokens/s
46.7
tokens/s
5.43
sec
pp2048+tg768
372
tokens/s
45.4
tokens/s
5.52
sec
pp1024+tg1024
416
tokens/s
47.9
tokens/s
2.48
sec
pp1280+tg3072
396
tokens/s
45.6
tokens/s
3.25
sec
pp384+tg1152
421
tokens/s
51.7
tokens/s
930
ms
pp64+tg1024
481
tokens/s
52.4
tokens/s
153
ms
pp16+tg1536
223
tokens/s
52.6
tokens/s
91
ms