TEST #2908 RESULTS
02/26/2026 - 3:47 PM
ACCELERATOR
MODEL
14.6
tokens/s
generation
9.65
sec
time to first token
134
tokens/s
prompt
59
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M1 4P+4E
RAM
8GB
OS
Darwin
Kernel Release
25.3.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.3.0: Wed Jan 28 20:53:01 PST 2026; root:xnu-12377.81.4~5/RELEASE_ARM64_T8103
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
147
tokens/s
15.7
tokens/s
7.04
sec
pp4096+tg256
131
tokens/s
11.7
tokens/s
31.43
sec
pp2048+tg256
142
tokens/s
13.8
tokens/s
14.51
sec
pp2048+tg768
142
tokens/s
13.6
tokens/s
14.52
sec
pp1024+tg1024
147
tokens/s
14.9
tokens/s
7.03
sec
pp1280+tg3072
146
tokens/s
13.1
tokens/s
8.86
sec
pp384+tg1152
150
tokens/s
15.8
tokens/s
2.62
sec
pp64+tg1024
141
tokens/s
16.4
tokens/s
512
ms
pp16+tg1536
60
tokens/s
16.1
tokens/s
327
ms