TEST #2926 RESULTS
02/27/2026 - 1:05 PM
ACCELERATOR
MODEL
22.3
tokens/s
generation
5.99
sec
time to first token
211
tokens/s
prompt
92
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M5 4P+6E
RAM
32GB
OS
Darwin
Kernel Release
25.2.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.2.0: Tue Nov 18 21:09:49 PST 2025; root:xnu-12377.61.12~1/RELEASE_ARM64_T8142
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
235
tokens/s
23.4
tokens/s
4.39
sec
pp4096+tg256
214
tokens/s
19.2
tokens/s
19.23
sec
pp2048+tg256
228
tokens/s
21.7
tokens/s
9.03
sec
pp2048+tg768
228
tokens/s
21.4
tokens/s
9.04
sec
pp1024+tg1024
235
tokens/s
22.6
tokens/s
4.41
sec
pp1280+tg3072
233
tokens/s
20.8
tokens/s
5.53
sec
pp384+tg1152
234
tokens/s
23.6
tokens/s
1.68
sec
pp64+tg1024
204
tokens/s
24.2
tokens/s
353
ms
pp16+tg1536
87
tokens/s
23.9
tokens/s
223
ms