TEST #3155 RESULTS
03/22/2026 - 8:45 PM
ACCELERATOR
MODEL
81.3
tokens/s
generation
367
ms
time to first token
3482
tokens/s
prompt
917
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 9 5900X 12-Core Processor (znver3)
RAM
127.9GB
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
4200
tokens/s
86.1
tokens/s
255
ms
pp4096+tg256
3369
tokens/s
78.0
tokens/s
1.23
sec
pp2048+tg256
3884
tokens/s
81.2
tokens/s
540
ms
pp2048+tg768
3957
tokens/s
79.6
tokens/s
530
ms
pp1024+tg1024
4268
tokens/s
82.1
tokens/s
252
ms
pp1280+tg3072
4147
tokens/s
80.6
tokens/s
320
ms
pp384+tg1152
4150
tokens/s
81.5
tokens/s
105
ms
pp64+tg1024
2694
tokens/s
81.5
tokens/s
36
ms
pp16+tg1536
667
tokens/s
81.5
tokens/s
35
ms