TEST #3059 RESULTS
03/11/2026 - 9:18 PM
ACCELERATOR
MODEL
57.1
tokens/s
generation
1.13
sec
time to first token
1203
tokens/s
prompt
394
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 3B Instruct - Q4_K - Small
SYSTEM
CPU
AMD Ryzen 7 5800X3D 8-Core Processor (znver3)
RAM
63.9GB
OS
Windows
Kernel Release
10.0
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64
RUNTIME
Name
llamafile
Version
0.9.3
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1484
tokens/s
57.7
tokens/s
704
ms
pp4096+tg256
1015
tokens/s
33.7
tokens/s
4.06
sec
pp2048+tg256
1305
tokens/s
49.7
tokens/s
1.59
sec
pp2048+tg768
1315
tokens/s
43.8
tokens/s
1.58
sec
pp1024+tg1024
1456
tokens/s
56.3
tokens/s
718
ms
pp1280+tg3072
1339
tokens/s
41.2
tokens/s
1.05
sec
pp384+tg1152
1505
tokens/s
69.2
tokens/s
266
ms
pp64+tg1024
504
tokens/s
84.8
tokens/s
136
ms
pp16+tg1536
905
tokens/s
77.7
tokens/s
27
ms