TEST #3266 RESULTS
04/09/2026 - 6:56 PM
ACCELERATOR
MODEL
35.7
tokens/s
generation
1.35
sec
time to first token
976
tokens/s
prompt
296
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 3 3100 4-Core Processor (znver2)
RAM
31.2GB
OS
Linux
Kernel Release
6.18.7-76061807-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #202601231045~1769703228~24.04~cb87b5b SMP PREEMPT_DYNAMIC Thu J
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1044
tokens/s
39.0
tokens/s
1.00
sec
pp4096+tg256
924
tokens/s
32.4
tokens/s
4.46
sec
pp2048+tg256
1039
tokens/s
35.8
tokens/s
2.00
sec
pp2048+tg768
1034
tokens/s
35.0
tokens/s
2.01
sec
pp1024+tg1024
1105
tokens/s
36.3
tokens/s
951
ms
pp1280+tg3072
1065
tokens/s
33.3
tokens/s
1.23
sec
pp384+tg1152
1140
tokens/s
36.3
tokens/s
361
ms
pp64+tg1024
1060
tokens/s
37.0
tokens/s
85
ms
pp16+tg1536
376
tokens/s
36.5
tokens/s
68
ms