You would expect to have high coverage, given the plasmids are short, so de novo assembly would be likely very easy. Given that each plasmid is present in different multiples, you would expect different coverage on each plasmid, so it might be best to approach it as a metagenome-type or transcriptome assembly, rather than a classic genome assembly.
Alternatively, Google search just told me there is a tool called plasmidSPADES designed for just your purpose. SPADES itself is a reliable genome assembler, so I would imagine plasmidSPADES to be a good choice. It does assume that there is also some whole bacterial genomes in the sequencing mix though.
As @terdon commented, it does depend on whether you have many identical copies of each plasmid, or many similar but not identical plasmids in the mix.