TEST #3351 RESULTS
04/19/2026 - 7:46 PM
ACCELERATOR
MODEL
31.2
tokens/s
generation
2.11
sec
time to first token
667
tokens/s
prompt
214
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Core i7-2600K CPU @ 3.40GHz (sandybridge)
RAM
23.4GB
OS
Linux
Kernel Release
5.15.0-171-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #181-Ubuntu SMP Fri Feb 6 22:44:50 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
764
tokens/s
39.3
tokens/s
1.37
sec
pp4096+tg256
553
tokens/s
12.4
tokens/s
7.48
sec
pp2048+tg256
666
tokens/s
22.5
tokens/s
3.12
sec
pp2048+tg768
679
tokens/s
20.2
tokens/s
3.06
sec
pp1024+tg1024
758
tokens/s
28.1
tokens/s
1.38
sec
pp1280+tg3072
681
tokens/s
17.1
tokens/s
1.91
sec
pp384+tg1152
772
tokens/s
39.9
tokens/s
512
ms
pp64+tg1024
741
tokens/s
55.2
tokens/s
96
ms
pp16+tg1536
388
tokens/s
45.9
tokens/s
49
ms