TEST #3625 RESULTS
05/26/2026 - 8:15 PM
ACCELERATOR
MODEL
367
tokens/s
generation
86
ms
time to first token
15400
tokens/s
prompt
4031
LocalScore
HOW YOU STACK UP
Explore All ResultsLlama 3.2 1B Instruct - Q4_K - Medium
SYSTEM
CPU
AMD EPYC 7B13 64-Core Processor (znver3)
RAM
125.6GB
OS
Linux
Kernel Release
6.8.0-117-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #117~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu May 7 22:17:46 UTC
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
19467
tokens/s
365
tokens/s
55
ms
pp4096+tg256
13863
tokens/s
314
tokens/s
299
ms
pp2048+tg256
17520
tokens/s
361
tokens/s
120
ms
pp2048+tg768
17511
tokens/s
355
tokens/s
120
ms
pp1024+tg1024
20210
tokens/s
375
tokens/s
53
ms
pp1280+tg3072
18832
tokens/s
344
tokens/s
71
ms
pp384+tg1152
21377
tokens/s
386
tokens/s
20
ms
pp64+tg1024
9210
tokens/s
403
tokens/s
9
ms
pp16+tg1536
606
tokens/s
397
tokens/s
29
ms