TEST #1840 RESULTS
11/21/2025 - 4:37 PM
ACCELERATOR
MODEL
45.4
tokens/s
generation
686
ms
time to first token
1881
tokens/s
prompt
499
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon Gold 5220R CPU @ 2.20GHz (cascadelake)
RAM
63.6GB
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
2293
tokens/s
45.6
tokens/s
470
ms
pp4096+tg256
1819
tokens/s
44.6
tokens/s
2.27
sec
pp2048+tg256
2095
tokens/s
45.3
tokens/s
1.00
sec
pp2048+tg768
2083
tokens/s
45.8
tokens/s
1.01
sec
pp1024+tg1024
2201
tokens/s
45.5
tokens/s
485
ms
pp1280+tg3072
2103
tokens/s
45.1
tokens/s
629
ms
pp384+tg1152
2296
tokens/s
45.4
tokens/s
189
ms
pp64+tg1024
1587
tokens/s
45.8
tokens/s
64
ms
pp16+tg1536
454
tokens/s
45.8
tokens/s
57
ms