TEST #3532 RESULTS
05/14/2026 - 2:35 PM
ACCELERATOR
MODEL
43.2
tokens/s
generation
3.68
sec
time to first token
336
tokens/s
prompt
158
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M5 Pro 5P+10E
RAM
24GB
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 19:33:50 PDT 2026; root:xnu-12377.101.15~1/RELEASE_ARM64_T6050
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
381
tokens/s
46.7
tokens/s
2.71
sec
pp4096+tg256
352
tokens/s
36.1
tokens/s
11.67
sec
pp2048+tg256
372
tokens/s
42.3
tokens/s
5.53
sec
pp2048+tg768
365
tokens/s
41.0
tokens/s
5.64
sec
pp1024+tg1024
375
tokens/s
44.5
tokens/s
2.75
sec
pp1280+tg3072
377
tokens/s
39.9
tokens/s
3.42
sec
pp384+tg1152
372
tokens/s
45.7
tokens/s
1.05
sec
pp64+tg1024
298
tokens/s
47.3
tokens/s
234
ms
pp16+tg1536
135
tokens/s
44.9
tokens/s
140
ms