TEST #3526 RESULTS
05/14/2026 - 10:31 AM
ACCELERATOR
MODEL
18.6
tokens/s
generation
5.30
sec
time to first token
330
tokens/s
prompt
105
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2650 v4 @ 2.20GHz (broadwell)
RAM
63.9GB
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
319
tokens/s
16.8
tokens/s
3.27
sec
pp4096+tg256
201
tokens/s
7.8
tokens/s
20.46
sec
pp2048+tg256
266
tokens/s
12.5
tokens/s
7.78
sec
pp2048+tg768
266
tokens/s
15.3
tokens/s
7.76
sec
pp1024+tg1024
320
tokens/s
18.4
tokens/s
3.25
sec
pp1280+tg3072
304
tokens/s
14.0
tokens/s
4.27
sec
pp384+tg1152
650
tokens/s
26.1
tokens/s
624
ms
pp64+tg1024
439
tokens/s
28.8
tokens/s
176
ms
pp16+tg1536
205
tokens/s
27.3
tokens/s
109
ms