TEST #134 RESULTS
03/28/2025 - 10:16 PM
ACCELERATOR
MODEL
104
tokens/s
generation
343
ms
time to first token
3739
tokens/s
prompt
1042
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon CPU E5-2690 v4 @ 2.60GHz (broadwell)
RAM
125.7GB
OS
Linux
Kernel Release
6.8.0-52-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #53~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Jan 15 19:18:46 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
4496
tokens/s
107
tokens/s
237
ms
pp4096+tg256
3613
tokens/s
93.3
tokens/s
1.14
sec
pp2048+tg256
4156
tokens/s
103
tokens/s
503
ms
pp2048+tg768
4152
tokens/s
102
tokens/s
503
ms
pp1024+tg1024
4468
tokens/s
106
tokens/s
239
ms
pp1280+tg3072
4339
tokens/s
98.9
tokens/s
304
ms
pp384+tg1152
4565
tokens/s
107
tokens/s
93
ms
pp64+tg1024
3151
tokens/s
110
tokens/s
29
ms
pp16+tg1536
713
tokens/s
109
tokens/s
31
ms