TEST #2050 RESULTS
12/27/2025 - 8:37 AM
ACCELERATOR
MODEL
24.1
tokens/s
generation
1.01
sec
time to first token
1510
tokens/s
prompt
331
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
RAM
119.7GB
OS
Linux
Kernel Release
6.14.0-1015-nvidia
Architecture
aarch64
Version
Cosmopolitan 3.9.7 MODE=aarch64; #15-Ubuntu SMP PREEMPT_DYNAMIC Tue Nov 25 18:02:16 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1625
tokens/s
23.1
tokens/s
675
ms
pp4096+tg256
1114
tokens/s
20.5
tokens/s
3.73
sec
pp2048+tg256
1413
tokens/s
22.5
tokens/s
1.50
sec
pp2048+tg768
1419
tokens/s
22.3
tokens/s
1.49
sec
pp1024+tg1024
1811
tokens/s
24.8
tokens/s
605
ms
pp1280+tg3072
1873
tokens/s
22.6
tokens/s
728
ms
pp384+tg1152
2522
tokens/s
24.6
tokens/s
180
ms
pp64+tg1024
1379
tokens/s
32.3
tokens/s
77
ms
pp16+tg1536
437
tokens/s
24.1
tokens/s
67
ms