PacBio sequencers produce two types of data. PacBio CCS (Circular Consensus Sequence) data is reasonably good quality and we recommend this for assembly in Geneious Prime. The Map to Reference algorithm in Geneious R9 handles these reads reasonably well, but for de novo assembly we recommend changing the advanced setting "Maximum Gap Size" to at least 5. Geneious R10 and later contains improvements that produce better results with PacBio CCS data.
PacBio CLR (Continuous Long Read), and similarly Oxford Nanopore Minion data, has longer reads but much lower quality. Because of the high error rate, these reads cannot be reliably assembled using the Geneious assembler or any other assembler Geneious has plugins for. However, different data sets tend to vary greatly in quality to the point where some data sets are quite usable for map to reference yet other data sets have quality so low that they are unusable in Geneious.
Note that Geneious cannot import Minion fast5 format. You can convert fast5 to fastq using Poretools (see http://poretools.readthedocs.org/en/latest/). Converted fastq data can then be imported into Geneious.