TEST #1790 RESULTS
11/13/2025 - 6:10 AM
ACCELERATOR
MODEL
62.7
tokens/s
generation
1.06
sec
time to first token
1119
tokens/s
prompt
405
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M3 Ultra 24P+8E
RAM
512GB
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 19:30:01 PDT 2025; root:xnu-12377.41.6~2/RELEASE_ARM64_T6031
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1356
tokens/s
68.3
tokens/s
770
ms
pp4096+tg256
1245
tokens/s
48.7
tokens/s
3.31
sec
pp2048+tg256
1316
tokens/s
59.6
tokens/s
1.57
sec
pp2048+tg768
1321
tokens/s
57.7
tokens/s
1.57
sec
pp1024+tg1024
1340
tokens/s
64.1
tokens/s
778
ms
pp1280+tg3072
1317
tokens/s
55.6
tokens/s
987
ms
pp384+tg1152
1273
tokens/s
68.7
tokens/s
315
ms
pp64+tg1024
706
tokens/s
72.0
tokens/s
104
ms
pp16+tg1536
199
tokens/s
69.9
tokens/s
93
ms