TEST #2280 RESULTS
01/13/2026 - 8:39 PM
ACCELERATOR
MODEL
38.3
tokens/s
generation
5.11
sec
time to first token
303
tokens/s
prompt
131
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC-Rome Processor (znver2)
RAM
30.6GB
OS
Linux
Kernel Release
6.12.63+deb13-cloud-amd64
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #1 SMP PREEMPT_DYNAMIC Debian 6.12.63-1 (2025-12-30)
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
309
tokens/s
44.3
tokens/s
3.33
sec
pp4096+tg256
203
tokens/s
32.9
tokens/s
20.18
sec
pp2048+tg256
293
tokens/s
37.2
tokens/s
7.01
sec
pp2048+tg768
299
tokens/s
35.3
tokens/s
6.86
sec
pp1024+tg1024
317
tokens/s
36.4
tokens/s
3.26
sec
pp1280+tg3072
326
tokens/s
35.1
tokens/s
3.96
sec
pp384+tg1152
346
tokens/s
40.3
tokens/s
1.13
sec
pp64+tg1024
385
tokens/s
41.7
tokens/s
183
ms
pp16+tg1536
253
tokens/s
41.5
tokens/s
84
ms