TEST #3139 RESULTS
03/20/2026 - 9:29 PM
ACCELERATOR
MODEL
30.9
tokens/s
generation
4.15
sec
time to first token
302
tokens/s
prompt
131
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M4 Pro 8P+4E
RAM
24GB
OS
Darwin
Kernel Release
25.3.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.3.0: Wed Jan 28 20:51:28 PST 2026; root:xnu-12377.91.3~2/RELEASE_ARM64_T6041
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
340
tokens/s
33.4
tokens/s
3.04
sec
pp4096+tg256
315
tokens/s
24.1
tokens/s
13.06
sec
pp2048+tg256
332
tokens/s
29.4
tokens/s
6.20
sec
pp2048+tg768
331
tokens/s
28.6
tokens/s
6.21
sec
pp1024+tg1024
339
tokens/s
31.6
tokens/s
3.05
sec
pp1280+tg3072
312
tokens/s
27.5
tokens/s
4.14
sec
pp384+tg1152
320
tokens/s
33.7
tokens/s
1.23
sec
pp64+tg1024
298
tokens/s
35.4
tokens/s
241
ms
pp16+tg1536
127
tokens/s
34.5
tokens/s
153
ms