TEST #1593 RESULTS
10/05/2025 - 7:38 PM
ACCELERATOR
MODEL
46.0
tokens/s
generation
2.07
sec
time to first token
583
tokens/s
prompt
235
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M3 Max 12P+4E
RAM
128GB
OS
Darwin
Kernel Release
25.0.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.0.0: Wed Sep 17 21:38:50 PDT 2025; root:xnu-12377.1.9~141/RELEASE_ARM64_T6031
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
710
tokens/s
50.3
tokens/s
1.46
sec
pp4096+tg256
650
tokens/s
35.2
tokens/s
6.33
sec
pp2048+tg256
688
tokens/s
43.6
tokens/s
3.00
sec
pp2048+tg768
679
tokens/s
42.2
tokens/s
3.04
sec
pp1024+tg1024
630
tokens/s
47.2
tokens/s
1.65
sec
pp1280+tg3072
561
tokens/s
40.5
tokens/s
2.31
sec
pp384+tg1152
653
tokens/s
50.7
tokens/s
607
ms
pp64+tg1024
495
tokens/s
53.1
tokens/s
147
ms
pp16+tg1536
180
tokens/s
51.4
tokens/s
106
ms