TEST #520 RESULTS
04/07/2025 - 8:19 PM
ACCELERATOR
MODEL
46.4
tokens/s
generation
2.44
sec
time to first token
484
tokens/s
prompt
210
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M2 Max 8P+4E
RAM
96GB
OS
Darwin
Kernel Release
23.5.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 23.5.0: Wed May 1 20:14:38 PDT 2024; root:xnu-10063.121.3~5/RELEASE_ARM64_T6020
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
571
tokens/s
50.6
tokens/s
1.81
sec
pp4096+tg256
539
tokens/s
35.7
tokens/s
7.63
sec
pp2048+tg256
563
tokens/s
43.6
tokens/s
3.66
sec
pp2048+tg768
564
tokens/s
42.6
tokens/s
3.65
sec
pp1024+tg1024
566
tokens/s
47.1
tokens/s
1.83
sec
pp1280+tg3072
562
tokens/s
41.0
tokens/s
2.30
sec
pp384+tg1152
537
tokens/s
51.1
tokens/s
734
ms
pp64+tg1024
338
tokens/s
53.8
tokens/s
207
ms
pp16+tg1536
116
tokens/s
52.3
tokens/s
155
ms