TEST #4174 RESULTS
09/17/2026 - 3:39 AM
ACCELERATOR
MODEL
52.6
tokens/s
generation
1.85
sec
time to first token
636
tokens/s
prompt
262
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M4 Max 12P+4E
RAM
48GB
OS
Darwin
Kernel Release
25.6.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.6.0: Tue Aug 18 17:48:46 PDT 2026; root:xnu-12377.161.15.700.19~2/RELEASE_ARM64_T6041
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
805
tokens/s
62.1
tokens/s
1.29
sec
pp4096+tg256
738
tokens/s
43.4
tokens/s
5.57
sec
pp2048+tg256
781
tokens/s
53.8
tokens/s
2.64
sec
pp2048+tg768
731
tokens/s
52.4
tokens/s
2.82
sec
pp1024+tg1024
675
tokens/s
58.3
tokens/s
1.54
sec
pp1280+tg3072
649
tokens/s
49.8
tokens/s
1.99
sec
pp384+tg1152
741
tokens/s
56.5
tokens/s
534
ms
pp64+tg1024
442
tokens/s
53.2
tokens/s
161
ms
pp16+tg1536
160
tokens/s
43.9
tokens/s
116
ms