is greatly slower in RF then in SD.Nextr

[gerroon]

Hi

FLUX schnell 2 steps takes like 50+ seconds in RF but it takes like 9 sec in SD.Next. Has anyone done any comparisons or speed tweaks to flux model workflow?

I am not sure if these are insanely different models.
RF flux1-schnell-fp8.safetensors
SD.Next FLUX.1-schnell

I am using with 3080 TI

[runew0lf]

so you're using a different model (and probably performance options on both, and theyre different speeds)....

[gerroon]

The thing is SD.Next downloads its own model and uses some kind of blob storage. I cant tell what I am looking at here. I am not sure which model would be the same one to download for RF.

I am not questioning RF abilities, it is more like how can I get to thatb kind of speed in RF, use the right model and settings.