TEST #3626 RESULTS
05/26/2026 - 8:22 PM
ACCELERATOR
MODEL
117
tokens/s
generation
325
ms
time to first token
3926
tokens/s
prompt
1122
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B 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
4456
tokens/s
120
tokens/s
238
ms
pp4096+tg256
3850
tokens/s
105
tokens/s
1.07
sec
pp2048+tg256
4381
tokens/s
116
tokens/s
476
ms
pp2048+tg768
4386
tokens/s
114
tokens/s
476
ms
pp1024+tg1024
4703
tokens/s
119
tokens/s
226
ms
pp1280+tg3072
4560
tokens/s
111
tokens/s
289
ms
pp384+tg1152
4904
tokens/s
120
tokens/s
86
ms
pp64+tg1024
3342
tokens/s
124
tokens/s
27
ms
pp16+tg1536
753
tokens/s
123
tokens/s
29
ms