Flux Dev Q5_K_M GGUF quantization (a nice balance of speed and quality in under 9 gigabytes) Reviews

great model, works sometime good.. but crashes a lot.. but that is my fault.. of course i try to get the max. ;) my rtx 3080 10GB Vram can handle it sometimes.. you need to try different settings. working most with 3-3.5 Guidens and 20-26 Steps.. sometimes with 2 batches.. with 1batch u are more on the stabile side. images size better not too high.. not over 1200px. and i can run it with 2 loras. cheers