TEST #3370 RESULTS
04/23/2026 - 1:41 PM
ACCELERATOR
MODEL
49.6
tokens/s
generation
2.05
sec
time to first token
583
tokens/s
prompt
242
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M4 Max 12P+4E
RAM
64GB
OS
Darwin
Kernel Release
25.4.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.4.0: Thu Mar 19 19:33:25 PDT 2026; root:xnu-12377.101.15~1/RELEASE_ARM64_T6041
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
812
tokens/s
61.2
tokens/s
1.28
sec
pp4096+tg256
672
tokens/s
41.8
tokens/s
6.12
sec
pp2048+tg256
654
tokens/s
50.8
tokens/s
3.15
sec
pp2048+tg768
639
tokens/s
49.1
tokens/s
3.22
sec
pp1024+tg1024
621
tokens/s
52.1
tokens/s
1.67
sec
pp1280+tg3072
614
tokens/s
45.5
tokens/s
2.10
sec
pp384+tg1152
654
tokens/s
49.6
tokens/s
605
ms
pp64+tg1024
419
tokens/s
49.0
tokens/s
170
ms
pp16+tg1536
166
tokens/s
47.5
tokens/s
112
ms