TEST #2780 RESULTS
02/22/2026 - 11:50 AM
ACCELERATOR
MODEL
48.6
tokens/s
generation
335
ms
time to first token
3827
tokens/s
prompt
822
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 9 9950X3D 16-Core Processor
RAM
60.3GB
OS
Linux
Kernel Release
6.17.0-14-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #14~24.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Jan 15 15:52:10 UTC 2
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
2815
tokens/s
48.5
tokens/s
384
ms
pp4096+tg256
3953
tokens/s
37.7
tokens/s
1.06
sec
pp2048+tg256
4876
tokens/s
48.0
tokens/s
442
ms
pp2048+tg768
4905
tokens/s
45.9
tokens/s
440
ms
pp1024+tg1024
5335
tokens/s
52.3
tokens/s
213
ms
pp1280+tg3072
4990
tokens/s
42.7
tokens/s
276
ms
pp384+tg1152
5250
tokens/s
54.0
tokens/s
92
ms
pp64+tg1024
1793
tokens/s
54.7
tokens/s
53
ms
pp16+tg1536
526
tokens/s
54.1
tokens/s
50
ms