TEST #4070 RESULTS
08/21/2026 - 9:14 PM
ACCELERATOR
MODEL
20.2
tokens/s
generation
29.47
sec
time to first token
48
tokens/s
prompt
32
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Embedded V1756B with Radeon Vega Gfx (znver2)
RAM
60.7GB
OS
Linux
Kernel Release
7.0.14-12-pve
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC PMX 7.0.14-12 (2026-08-11T11:05Z)
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
61
tokens/s
22.7
tokens/s
16.91
sec
pp4096+tg256
43
tokens/s
16.2
tokens/s
96.13
sec
pp2048+tg256
45
tokens/s
19.5
tokens/s
45.45
sec
pp2048+tg768
44
tokens/s
18.7
tokens/s
46.59
sec
pp1024+tg1024
46
tokens/s
20.4
tokens/s
22.32
sec
pp1280+tg3072
46
tokens/s
17.5
tokens/s
28.17
sec
pp384+tg1152
48
tokens/s
21.9
tokens/s
8.00
sec
pp64+tg1024
52
tokens/s
22.8
tokens/s
1.26
sec
pp16+tg1536
48
tokens/s
22.3
tokens/s
378
ms