TEST #3071 RESULTS
03/13/2026 - 5:59 PM
ACCELERATOR
MODEL
48.3
tokens/s
generation
2.13
sec
time to first token
580
tokens/s
prompt
236
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M4 Max 10P+4E
RAM
36GB
OS
Darwin
Kernel Release
25.1.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.1.0: Mon Oct 20 19:34:05 PDT 2025; root:xnu-12377.41.6~2/RELEASE_ARM64_T6041
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
664
tokens/s
52.2
tokens/s
1.56
sec
pp4096+tg256
608
tokens/s
38.0
tokens/s
6.77
sec
pp2048+tg256
645
tokens/s
45.8
tokens/s
3.20
sec
pp2048+tg768
645
tokens/s
44.8
tokens/s
3.19
sec
pp1024+tg1024
662
tokens/s
49.5
tokens/s
1.57
sec
pp1280+tg3072
651
tokens/s
43.2
tokens/s
1.99
sec
pp384+tg1152
649
tokens/s
52.7
tokens/s
610
ms
pp64+tg1024
504
tokens/s
55.0
tokens/s
144
ms
pp16+tg1536
187
tokens/s
53.7
tokens/s
102
ms