TEST #3561 RESULTS
05/17/2026 - 7:37 PM
ACCELERATOR
MODEL
22.2
tokens/s
generation
5.55
sec
time to first token
230
tokens/s
prompt
97
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M3 Pro 5P+6E
RAM
18GB
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 19:31:56 PDT 2026; root:xnu-12377.101.15~1/RELEASE_ARM64_T6030
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
256
tokens/s
23.6
tokens/s
4.04
sec
pp4096+tg256
227
tokens/s
18.3
tokens/s
18.06
sec
pp2048+tg256
247
tokens/s
21.5
tokens/s
8.34
sec
pp2048+tg768
247
tokens/s
21.0
tokens/s
8.33
sec
pp1024+tg1024
255
tokens/s
22.6
tokens/s
4.05
sec
pp1280+tg3072
253
tokens/s
20.3
tokens/s
5.09
sec
pp384+tg1152
257
tokens/s
23.9
tokens/s
1.53
sec
pp64+tg1024
230
tokens/s
24.6
tokens/s
318
ms
pp16+tg1536
99
tokens/s
24.2
tokens/s
201
ms