TEST #2344 RESULTS
01/21/2026 - 6:25 PM
ACCELERATOR
MODEL
42.7
tokens/s
generation
1.31
sec
time to first token
1082
tokens/s
prompt
328
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
ARMv8 Processor rev 1 (v8l)
RAM
7.4GB
OS
Linux
Kernel Release
5.15.148-tegra
Architecture
aarch64
Version
Cosmopolitan 3.9.7 MODE=aarch64; #1 SMP PREEMPT Thu Sep 18 15:08:33 PDT 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1321
tokens/s
45.5
tokens/s
796
ms
pp4096+tg256
850
tokens/s
34.9
tokens/s
4.85
sec
pp2048+tg256
1128
tokens/s
42.1
tokens/s
1.84
sec
pp2048+tg768
1117
tokens/s
40.9
tokens/s
1.86
sec
pp1024+tg1024
1321
tokens/s
43.7
tokens/s
795
ms
pp1280+tg3072
1261
tokens/s
40.5
tokens/s
1.04
sec
pp384+tg1152
1494
tokens/s
45.8
tokens/s
276
ms
pp64+tg1024
1162
tokens/s
46.8
tokens/s
74
ms
pp16+tg1536
80
tokens/s
44.2
tokens/s
225
ms