The only con of using plasmid for sequencing is the need to ligate and culture the bacteria. Other than that, it's less contaminated, enables longer sequencing reads and uses the same set of sequencing primers for each insert. I use it when there are problems with direct PCR - like unwanted products or unspecific primers.

- From plasmid:

• Pros: Can get complete sequence of target dna, clean sequencing result, use universal sequencing primers on plasmid backbone.
• Cons: time consuming (gel elution, ligation, transformation, culture, plasmid isolation ~ 2 working days), mutation will show up if you use normal taq polymerase.

- From direct PCR:

• Pros: simple and fast, you need enzyme treat (exonuclease and SAP) to remove remained dNTP and primer before sequencing, mutations not show even use cheap taq, if PCR have unspecific product, other sequencing primers (nested) is necessary.
• Cons: can not read complete of target (loss ~ 50 bp in beginning) in a single read. 

Tip: In my lab, after cloning using bacteria, we use colony PCR technique and direct sequecing to reduce the time. ^^

From direct PCR, the sequence can be ambiguous, allelic diversity, and usually represents the the most amplified product.

From plasmid, 1 sequence = 1 allele and you can detect different alleles by sequencing different transformants

Plasmid sequencing may not show heterozygous mutations whereas direct sequencing does detect it. Mutations can be in both plasmid based or direct sequencing if you use ordinary Taq polymerase. For sequencing applications always use proof-reading enzymes such as Pfu.

We use plasmid sequencing when we need to detect different alleles that appear in heterozygous in direct PCR sequence. Plasmid sequencing is very most useful in the case of deletions or insertions to see the two alleles.