A Lora that lets Flux generate usable images in as few as 8 sampling steps. The compressed step count trades off some quality/detail, which makes the Model Sampling Flux node's Max Shift/Base Shift controls especially valuable as a way to claw back detail without adding steps back.
Load the Turbo Lora with the Flux model for fast generation, then use Model Sampling Flux's Max Shift/Base Shift to fix or improve detail issues the compressed step count introduces, instead of reverting to a slower non-Turbo workflow.