TEST #1535 RESULTS
09/21/2025 - 2:37 PM
ACCELERATOR
MODEL
38.8
tokens/s
generation
1.43
sec
time to first token
889
tokens/s
prompt
289
LocalScore
HOW YOU STACK UP
Explore All ResultsMeta Llama 3.1 8B Instruct - Q4_K - Medium
SYSTEM
CPU
Intel Xeon Gold 6254 CPU @ 3.10GHz (cascadelake)
RAM
251.5GB
OS
Linux
Kernel Release
6.8.0-83-generic
Architecture
x86_64
Version
Cosmopolitan 3.9.7 MODE=x86_64; #83-Ubuntu SMP PREEMPT_DYNAMIC Fri Sep 5 21:46:54 UTC 2025
RUNTIME
Name
llamafile
Version
0.9.2
Commit Hash
a30b324
DETAILED RESULTS
TEST NAME
PROMPT
GENERATION
TTFT
pp1024+tg16
1034
tokens/s
41.0
tokens/s
1.02
sec
pp4096+tg256
881
tokens/s
31.9
tokens/s
4.68
sec
pp2048+tg256
974
tokens/s
37.3
tokens/s
2.13
sec
pp2048+tg768
974
tokens/s
36.5
tokens/s
2.13
sec
pp1024+tg1024
1033
tokens/s
39.4
tokens/s
1.02
sec
pp1280+tg3072
1017
tokens/s
35.6
tokens/s
1.28
sec
pp384+tg1152
1082
tokens/s
41.7
tokens/s
378
ms
pp64+tg1024
853
tokens/s
43.3
tokens/s
97
ms
pp16+tg1536
155
tokens/s
42.5
tokens/s
125
ms