TEST #2984 RESULTS
03/05/2026 - 10:54 PM
ACCELERATOR
MODEL
11.5
tokens/s
generation
11.35
sec
time to first token
122
tokens/s
prompt
50
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 9 9950X 16-Core Processor
RAM
31.2GB
OS
Windows
Kernel Release
10.0
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64
RUNTIME
Name
llamafile
Version
0.9.3
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
125
tokens/s
11.8
tokens/s
8.25
sec
pp4096+tg256
111
tokens/s
10.5
tokens/s
36.93
sec
pp2048+tg256
119
tokens/s
11.1
tokens/s
17.29
sec
pp2048+tg768
118
tokens/s
11.0
tokens/s
17.40
sec
pp1024+tg1024
124
tokens/s
11.9
tokens/s
8.36
sec
pp1280+tg3072
129
tokens/s
11.1
tokens/s
10.04
sec
pp384+tg1152
128
tokens/s
11.9
tokens/s
3.08
sec
pp64+tg1024
131
tokens/s
12.1
tokens/s
569
ms
pp16+tg1536
116
tokens/s
12.0
tokens/s
220
ms