I've been using the Polar CS600X with Power for a few years now - this is something of a long-term test report.

1. Accuracy.  I am unable to comment on accuracy, not having another power meter with which to compare it's reported power values.  That being said, my experiences would indicate that it's pretty consistent, and since I am just using the unit to evaluate where I am in my training programme rather than accurately measuring power, it does the job OK (I refer to Polar Watts!).
2. Reliability.  I am afraid that the unit loses marks here.  To keep the Power sensor working involves (a) correct positioning of the sensor on the chainstay, (b) a good electrical connection with the seatstay mounted battery pack, (c) a good electrical connection with the chain movement detector and (d) keeping the crank arm magnet alined correctly with the appropriate part of the power sensor. In my view you need to be prepared for a fair bit of adjustment and tinkering to keep things working.  I've just spent quite a few hours over several days making the device work again: it turned out to be a poor connection with the chain speed sensor.  This on a bike which is mounted on a turbo in the garage.
3. The GPS sensor.  This is so cryptic that making it communicate properly with the main unit is confused and confusing.

I have to say that in comparison with the Garmin 500, the Polar loses out in the usability stakes.  When it works, it's just fine, but solving why the power meter is failing to report power or cadence to the CS600X can be challenging.

Over the last few days, neither cadence nor power were being reported: I initially thought this was a failure in the cadence sensor, but actually it turned out to be a connection problem with the chain speed sensor.  Quite why that would prevent cadence from showing up on the computer is a mystery to me.