TEST #4141 RESULTS
09/09/2026 - 6:49 PM
ACCELERATOR
MODEL
41.2
tokens/s
generation
137
ms
time to first token
9866
tokens/s
prompt
1439
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD Ryzen Threadripper 7960X 24-Cores (znver4)
RAM
94.3GB
OS
Linux
Kernel Release
6.18.40.1-microsoft-standard-WSL2
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC Fri Jul 31 22:12:15 UTC 2026
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
12681
tokens/s
41.1
tokens/s
103
ms
pp4096+tg256
12263
tokens/s
40.4
tokens/s
357
ms
pp2048+tg256
13059
tokens/s
41.1
tokens/s
180
ms
pp2048+tg768
13186
tokens/s
40.2
tokens/s
179
ms
pp1024+tg1024
13463
tokens/s
40.7
tokens/s
101
ms
pp1280+tg3072
11594
tokens/s
40.4
tokens/s
136
ms
pp384+tg1152
10500
tokens/s
42.0
tokens/s
60
ms
pp64+tg1024
1609
tokens/s
43.0
tokens/s
58
ms
pp16+tg1536
444
tokens/s
42.1
tokens/s
54
ms