TEST #644 RESULTS
04/09/2025 - 4:55 PM
ACCELERATOR
MODEL
96.6
tokens/s
generation
93
ms
time to first token
14632
tokens/s
prompt
2481
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon Platinum 8468 (sapphirerapids)
RAM
196.6GB
OS
Linux
Kernel Release
5.15.0-133-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #144-Ubuntu SMP Fri Feb 7 20:47:38 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
19810
tokens/s
97.4
tokens/s
62
ms
pp4096+tg256
14512
tokens/s
88.6
tokens/s
309
ms
pp2048+tg256
19180
tokens/s
95.9
tokens/s
118
ms
pp2048+tg768
19165
tokens/s
95.4
tokens/s
117
ms
pp1024+tg1024
19834
tokens/s
95.3
tokens/s
62
ms
pp1280+tg3072
17655
tokens/s
93.3
tokens/s
83
ms
pp384+tg1152
16686
tokens/s
98.3
tokens/s
32
ms
pp64+tg1024
3792
tokens/s
104
tokens/s
25
ms
pp16+tg1536
1057
tokens/s
101
tokens/s
24
ms