TEST #3251 RESULTS
04/07/2026 - 1:51 PM
ACCELERATOR
MODEL
19.5
tokens/s
generation
2.62
sec
time to first token
547
tokens/s
prompt
160
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Core i7-10750H CPU @ 2.60GHz (skylake)
RAM
39.8GB
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
612
tokens/s
22.2
tokens/s
1.72
sec
pp4096+tg256
448
tokens/s
7.3
tokens/s
9.28
sec
pp2048+tg256
535
tokens/s
13.3
tokens/s
3.91
sec
pp2048+tg768
534
tokens/s
12.1
tokens/s
3.91
sec
pp1024+tg1024
602
tokens/s
17.9
tokens/s
1.74
sec
pp1280+tg3072
588
tokens/s
10.7
tokens/s
2.23
sec
pp384+tg1152
654
tokens/s
26.0
tokens/s
610
ms
pp64+tg1024
671
tokens/s
36.6
tokens/s
108
ms
pp16+tg1536
278
tokens/s
29.9
tokens/s
68
ms