TEST #3124 RESULTS
03/19/2026 - 10:44 PM
ACCELERATOR
MODEL
13.7
tokens/s
generation
10.07
sec
time to first token
125
tokens/s
prompt
55
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M3 4P+4E
RAM
16GB
OS
Darwin
Kernel Release
25.4.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.4.0: Thu Mar 5 22:46:43 PST 2026; root:xnu-12377.101.14~19/RELEASE_ARM64_T8122
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
143
tokens/s
15.5
tokens/s
7.22
sec
pp4096+tg256
128
tokens/s
12.1
tokens/s
31.98
sec
pp2048+tg256
139
tokens/s
14.1
tokens/s
14.85
sec
pp2048+tg768
137
tokens/s
13.9
tokens/s
15.07
sec
pp1024+tg1024
135
tokens/s
14.8
tokens/s
7.64
sec
pp1280+tg3072
128
tokens/s
12.5
tokens/s
10.09
sec
pp384+tg1152
137
tokens/s
13.4
tokens/s
2.87
sec
pp64+tg1024
121
tokens/s
13.7
tokens/s
591
ms
pp16+tg1536
57
tokens/s
13.4
tokens/s
344
ms