TEST #3320 RESULTS
04/15/2026 - 10:43 PM
ACCELERATOR
MODEL
101
tokens/s
generation
383
ms
time to first token
3307
tokens/s
prompt
955
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
12th Gen Intel Core i9-12900K (alderlake)
RAM
91.8GB
OS
Linux
Kernel Release
6.8.0-107-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #107-Ubuntu SMP PREEMPT_DYNAMIC Fri Mar 13 19:51:50 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
3642
tokens/s
103
tokens/s
291
ms
pp4096+tg256
3259
tokens/s
90.4
tokens/s
1.27
sec
pp2048+tg256
3713
tokens/s
99.9
tokens/s
562
ms
pp2048+tg768
3763
tokens/s
98.8
tokens/s
554
ms
pp1024+tg1024
3990
tokens/s
103
tokens/s
266
ms
pp1280+tg3072
3945
tokens/s
96.2
tokens/s
334
ms
pp384+tg1152
3865
tokens/s
104
tokens/s
109
ms
pp64+tg1024
2732
tokens/s
107
tokens/s
32
ms
pp16+tg1536
853
tokens/s
105
tokens/s
27
ms