TEST #3451 RESULTS
05/05/2026 - 9:04 PM
ACCELERATOR
MODEL
40.5
tokens/s
generation
11.22
sec
time to first token
121
tokens/s
prompt
76
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Core Ultra 5 238V (arrowlake-s)
RAM
31.6GB
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
129
tokens/s
47.0
tokens/s
7.98
sec
pp4096+tg256
110
tokens/s
35.9
tokens/s
37.11
sec
pp2048+tg256
123
tokens/s
43.7
tokens/s
16.64
sec
pp2048+tg768
116
tokens/s
36.6
tokens/s
17.70
sec
pp1024+tg1024
129
tokens/s
43.0
tokens/s
7.98
sec
pp1280+tg3072
131
tokens/s
35.4
tokens/s
9.80
sec
pp384+tg1152
127
tokens/s
42.4
tokens/s
3.05
sec
pp64+tg1024
113
tokens/s
38.9
tokens/s
592
ms
pp16+tg1536
116
tokens/s
42.0
tokens/s
157
ms