TEST #1630 RESULTS
10/10/2025 - 12:44 PM
ACCELERATOR
MODEL
110
tokens/s
generation
73
ms
time to first token
17936
tokens/s
prompt
2996
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 9455P 48-Core Processor
RAM
755.1GB
OS
Linux
Kernel Release
6.8.0-71-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #71-Ubuntu SMP PREEMPT_DYNAMIC Tue Jul 22 16:52:38 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
24666
tokens/s
121
tokens/s
50
ms
pp4096+tg256
19998
tokens/s
98.6
tokens/s
215
ms
pp2048+tg256
22076
tokens/s
99.9
tokens/s
103
ms
pp2048+tg768
22034
tokens/s
99.6
tokens/s
103
ms
pp1024+tg1024
24643
tokens/s
106
tokens/s
50
ms
pp1280+tg3072
21723
tokens/s
99.9
tokens/s
68
ms
pp384+tg1152
20727
tokens/s
116
tokens/s
26
ms
pp64+tg1024
4367
tokens/s
125
tokens/s
22
ms
pp16+tg1536
1187
tokens/s
121
tokens/s
20
ms