TEST #3597 RESULTS
05/22/2026 - 2:15 PM
ACCELERATOR
MODEL
66.3
tokens/s
generation
307
ms
time to first token
4258
tokens/s
prompt
972
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 7 5700X 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
0.9.3
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
5432
tokens/s
69.2
tokens/s
203
ms
pp4096+tg256
4006
tokens/s
68.1
tokens/s
1.04
sec
pp2048+tg256
4601
tokens/s
69.1
tokens/s
463
ms
pp2048+tg768
4863
tokens/s
67.7
tokens/s
436
ms
pp1024+tg1024
5419
tokens/s
65.5
tokens/s
203
ms
pp1280+tg3072
5147
tokens/s
64.7
tokens/s
263
ms
pp384+tg1152
5790
tokens/s
61.0
tokens/s
82
ms
pp64+tg1024
2280
tokens/s
66.0
tokens/s
43
ms
pp16+tg1536
787
tokens/s
65.1
tokens/s
36
ms