TEST #1619 RESULTS
10/08/2025 - 9:03 PM
ACCELERATOR
MODEL
54.2
tokens/s
generation
1.71
sec
time to first token
660
tokens/s
prompt
275
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M1 Ultra 16P+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 21:41:45 PDT 2025; root:xnu-12377.1.9~141/RELEASE_ARM64_T6000
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
815
tokens/s
57.2
tokens/s
1.27
sec
pp4096+tg256
791
tokens/s
44.3
tokens/s
5.20
sec
pp2048+tg256
817
tokens/s
51.8
tokens/s
2.53
sec
pp2048+tg768
825
tokens/s
50.6
tokens/s
2.50
sec
pp1024+tg1024
807
tokens/s
54.8
tokens/s
1.29
sec
pp1280+tg3072
789
tokens/s
49.0
tokens/s
1.64
sec
pp384+tg1152
694
tokens/s
58.4
tokens/s
570
ms
pp64+tg1024
321
tokens/s
61.3
tokens/s
215
ms
pp16+tg1536
82
tokens/s
60.1
tokens/s
211
ms