Which file is the model file that needs to be downloaded?

[KLL535]

The instructions are completely unclear
Which file is the model file that needs to be downloaded? What is the difference between these strange letters and numbers?
flux1-dev-F16.gguf
flux1-dev-Q2_K.gguf
flux1-dev-Q3_K_S.gguf
flux1-dev-Q4_0.gguf
flux1-dev-Q4_1.gguf
flux1-dev-Q4_K_S.gguf
flux1-dev-Q5_0.gguf
flux1-dev-Q5_1.gguf
flux1-dev-Q5_K_S.gguf
flux1-dev-Q6_K.gguf
flux1-dev-Q8_0.gguf

"LoRA loading is experimental"
What does this mean? Will Lora work with these models or not?

[KLL535]

flux1-dev-Q4_0.gguf
Generation time = 109.34s

flux_dev.safetensors
https://civitai.com/models/618692/flux
Generation time = 90.62s

What's the point?
In forge the nf4 model give a real speed boost, but this model works even slower than the standard one.

[al-swaiti]

what is ur gpu ?
as ur gpu bigger go with bigger size version (choose gguf which have ur gpu size ) i have 12g so i prefer q8 and q6-k

[KLL535]

what is ur gpu ? as ur gpu bigger go with bigger size version (choose gguf which have ur gpu size ) i have 12g so i prefer q8 and q6-k

I have RTX3060 12Gb.
I tested these models and I didn't like them at all. In theory, they are smaller and should work faster? No, they work even slower.
I choose the usual flux-dev-fp8
flux1-dev-Q8_0.gguf - the same size. So what's the point?