TEST #3161 RESULTS
03/23/2026 - 10:54 AM
ACCELERATOR
MODEL
54.7
tokens/s
generation
848
ms
time to first token
1525
tokens/s
prompt
462
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Celeron G4900 CPU @ 3.10GHz (skylake)
RAM
7.6GB
OS
Linux
Kernel Release
5.15.0-173-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #183-Ubuntu SMP Fri Mar 6 13:29:34 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1875
tokens/s
54.7
tokens/s
564
ms
pp4096+tg256
1395
tokens/s
39.0
tokens/s
2.96
sec
pp2048+tg256
1690
tokens/s
51.0
tokens/s
1.23
sec
pp2048+tg768
1689
tokens/s
49.3
tokens/s
1.23
sec
pp1024+tg1024
1860
tokens/s
57.0
tokens/s
567
ms
pp1280+tg3072
1778
tokens/s
46.4
tokens/s
737
ms
pp384+tg1152
1904
tokens/s
62.8
tokens/s
216
ms
pp64+tg1024
1134
tokens/s
67.7
tokens/s
70
ms
pp16+tg1536
405
tokens/s
64.8
tokens/s
53
ms