TEST #2922 RESULTS
02/27/2026 - 12:32 PM
ACCELERATOR
MODEL
38.1
tokens/s
generation
3.12
sec
time to first token
384
tokens/s
prompt
167
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M1 Max 8P+2E
RAM
64GB
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 21:09:40 PST 2025; root:xnu-12377.61.12~1/RELEASE_ARM64_T6000
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
441
tokens/s
41.5
tokens/s
2.35
sec
pp4096+tg256
420
tokens/s
30.4
tokens/s
9.78
sec
pp2048+tg256
441
tokens/s
36.7
tokens/s
4.68
sec
pp2048+tg768
442
tokens/s
35.7
tokens/s
4.66
sec
pp1024+tg1024
444
tokens/s
39.2
tokens/s
2.33
sec
pp1280+tg3072
442
tokens/s
34.1
tokens/s
2.92
sec
pp384+tg1152
431
tokens/s
40.9
tokens/s
914
ms
pp64+tg1024
282
tokens/s
42.6
tokens/s
250
ms
pp16+tg1536
109
tokens/s
41.7
tokens/s
169
ms