TEST #3082 RESULTS
03/15/2026 - 12:58 PM
ACCELERATOR
MODEL
16.2
tokens/s
generation
912
ms
time to first token
1646
tokens/s
prompt
308
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD RYZEN AI MAX+ 395 w/ Radeon 8060S
RAM
62.4GB
OS
Linux
Kernel Release
6.17.0-1012-oem
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #12-Ubuntu SMP PREEMPT_DYNAMIC Tue Feb 10 04:51:46 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.3
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2022
tokens/s
17.0
tokens/s
565
ms
pp4096+tg256
1281
tokens/s
4.5
tokens/s
3.41
sec
pp2048+tg256
1697
tokens/s
11.6
tokens/s
1.31
sec
pp2048+tg768
1702
tokens/s
10.5
tokens/s
1.30
sec
pp1024+tg1024
2017
tokens/s
15.4
tokens/s
567
ms
pp1280+tg3072
1918
tokens/s
8.7
tokens/s
738
ms
pp384+tg1152
2721
tokens/s
22.1
tokens/s
172
ms
pp64+tg1024
968
tokens/s
31.0
tokens/s
85
ms
pp16+tg1536
486
tokens/s
25.3
tokens/s
49
ms