TEST #2552 RESULTS
02/07/2026 - 2:52 AM
ACCELERATOR
MODEL
39.8
tokens/s
generation
5.74
sec
time to first token
254
tokens/s
prompt
121
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen 7 3700X 8-Core Processor (znver2)
RAM
31.3GB
OS
Windows
Kernel Release
10.0
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
257
tokens/s
45.1
tokens/s
4.01
sec
pp4096+tg256
205
tokens/s
33.8
tokens/s
20.04
sec
pp2048+tg256
243
tokens/s
39.4
tokens/s
8.47
sec
pp2048+tg768
243
tokens/s
38.7
tokens/s
8.47
sec
pp1024+tg1024
258
tokens/s
41.1
tokens/s
3.98
sec
pp1280+tg3072
260
tokens/s
36.0
tokens/s
4.96
sec
pp384+tg1152
277
tokens/s
41.9
tokens/s
1.41
sec
pp64+tg1024
295
tokens/s
41.7
tokens/s
249
ms
pp16+tg1536
249
tokens/s
40.4
tokens/s
80
ms