TEST #1193 RESULTS
07/08/2025 - 5:14 PM
ACCELERATOR
MODEL
37.9
tokens/s
generation
3.10
sec
time to first token
385
tokens/s
prompt
168
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M1 Max 8P+2E
RAM
32GB
OS
Darwin
Kernel Release
24.2.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 24.2.0: Fri Dec 6 19:01:59 PST 2024; root:xnu-11215.61.5~2/RELEASE_ARM64_T6000
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
450
tokens/s
40.9
tokens/s
2.30
sec
pp4096+tg256
422
tokens/s
30.1
tokens/s
9.74
sec
pp2048+tg256
443
tokens/s
36.3
tokens/s
4.65
sec
pp2048+tg768
444
tokens/s
35.4
tokens/s
4.64
sec
pp1024+tg1024
448
tokens/s
38.6
tokens/s
2.31
sec
pp1280+tg3072
443
tokens/s
34.1
tokens/s
2.92
sec
pp384+tg1152
429
tokens/s
41.2
tokens/s
917
ms
pp64+tg1024
290
tokens/s
42.9
tokens/s
243
ms
pp16+tg1536
95
tokens/s
41.9
tokens/s
191
ms