One of the ways to do it is with seqkit:

seqkit seq -M 10000 -m 500 file.fas > new_file.fas

$ bioawk -c fastx '{ml=500;ML=10000;print (length($seq)>ml && length($seq)<ML)? (">"$name"\n"$seq) :""}' test.fna
$ cutadapt --quiet -m 500 -M 10000 test.fna

Using BBMap suite:

reformat.sh in=input.fa out=filterd.fa minlength=500 maxlength=10000