In the middle of something along these lines, but only with a focus on 454 data. The company I work with has been using NextGene for some time, and when we surveyed the market for a suitable package for their work (variant detection in amplicon sequencing) it stacked up very well. They're very happy with the output and have no problems validating their variants. I can't comment on it for other applications or sequencing technologies, and have only driven it in 'test mode' a few times.

Why not give it a try, and significantly reduce your bioinformatic support cost

I think we, Bioinformatics people, are very lucky as the best programs for most of our task are free. It's quite unusual situation. My friend worked in Chemoinformatics and there situation is completely opposite. All standard, widely-used tools are commercial, and even academic users have to pay!

Answering your question, yes I think the best currently available solutions are those free solutions you mentioned. Free software usually is widely tested. The code is open and often published, so you know in details how it works. On the other, commercial tools often have got nice GUI and very good support. It's very important especially for inexperienced users. But, free solutions usually evolve faster (anyone can participate) and are highly configurable (or you even can rewrite some parts;) ). In addition, it's free software that is mostly cited in NGS-related publications.

Have a look at this thread as it's similar question.