TEST #1005 RESULTS
06/06/2025 - 5:12 AM
ACCELERATOR
MODEL
38.9
tokens/s
generation
2.36
sec
time to first token
524
tokens/s
prompt
205
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 7 5700X3D 8-Core Processor (znver3)
RAM
31.9GB
OS
Windows
Kernel Release
10.0
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
661
tokens/s
44.6
tokens/s
1.57
sec
pp4096+tg256
511
tokens/s
22.0
tokens/s
8.05
sec
pp2048+tg256
602
tokens/s
34.1
tokens/s
3.43
sec
pp2048+tg768
601
tokens/s
32.2
tokens/s
3.44
sec
pp1024+tg1024
648
tokens/s
39.4
tokens/s
1.60
sec
pp1280+tg3072
624
tokens/s
29.6
tokens/s
2.08
sec
pp384+tg1152
608
tokens/s
45.8
tokens/s
650
ms
pp64+tg1024
207
tokens/s
52.9
tokens/s
325
ms
pp16+tg1536
257
tokens/s
49.4
tokens/s
83
ms