TEST #2808 RESULTS
02/23/2026 - 7:50 AM
ACCELERATOR
MODEL
23.6
tokens/s
generation
5.73
sec
time to first token
219
tokens/s
prompt
97
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Apple M2 Pro 6P+4E
RAM
16GB
OS
Darwin
Kernel Release
25.4.0
Architecture
arm64
Version
Cosmopolitan 3.9.7 MODE=aarch64; Darwin Kernel Version 25.4.0: Sun Feb 8 23:32:59 PST 2026; root:xnu-12377.100.591.501.2~2/RELEASE_ARM64_T6020
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
243
tokens/s
25.6
tokens/s
4.25
sec
pp4096+tg256
225
tokens/s
18.8
tokens/s
18.29
sec
pp2048+tg256
238
tokens/s
22.8
tokens/s
8.63
sec
pp2048+tg768
238
tokens/s
21.9
tokens/s
8.66
sec
pp1024+tg1024
244
tokens/s
23.6
tokens/s
4.24
sec
pp1280+tg3072
242
tokens/s
21.1
tokens/s
5.33
sec
pp384+tg1152
238
tokens/s
25.6
tokens/s
1.65
sec
pp64+tg1024
213
tokens/s
26.8
tokens/s
336
ms
pp16+tg1536
91
tokens/s
26.4
tokens/s
211
ms