Class A to XX Watts

I've only used pure Class A (Aleph J), never an amp that started out in Class A for the first few watts only. Wouldn't the transition out of Class A be audible? Seems to me that if it is then that would be a bad thing. If it isn't audible, then why would anyone design it that way to begin with (marketing?)?
Post removed 
Good 'compromise' and one heck of an amp would be the Pass 'a' series.
The XA30.5 is fully 30x2 in 'a'. BUT it redlines at around 3x that+ when firmly in A/B.
Unless you are a panel Elizabeth, me and many others, this is a decent amount of power. I suspect the XA60.5 would be the last amp I would ever need....and never leave 'a'.

Tough to beat a hi-headroom 'a' amp.
Just how noticeable the transition from Class A to Class B depends on different conditions. Often times the transition will happen as either bass demands and/or overall volume increases. During such times ones ears might not be as sensitive to the change from Class A to Class B. Usually the longer an amp can maintain Class A operation, the longer the amp can offer it's contribution to better sound, but the longer an amp can maintain Class A, the more expensive and less efficient the amp becomes. Better to to switch to Class B than to run out of Class A Watts, and run into distortion and clipping. High biased Class AB amps can be an appropriate combination of sound quality, efficiency and economy.
Unsound, that sounds very sound, and one of the benefits go higher efficiency speakers
Pubul57, unto itself that might appear to be the case, but speakers are another issue unto themselves. The net result of the interaction between them is not necessarily as it might so appear. Speakers that are less efficient aren't so for the sake of it, but for advantages elsewhere.
very true. ahh, the tradeoffs. but i do think the A/B tradeoff maybe, as you, well worth it in a given system.
Depending on the speakers and listening levels a high biased AB amplifier may never switch out of class A.
Class A biasing is beneficial when it comes to sound quality, but there are many, many other things that it affects, most notably reliability: constant thermal cycling leads to failures. Biasing is a thin veneer spec that has just a fraction to do with the overall sound and character of an amplifier - it's a detail, not a foundation.

My Boulder 1060 operates up to 17 watts in Class A before switching to Class AB, however Boulder spent a lot of time perfecting the circuit design & notching out crossover distortion. The Rowland 625 for example runs in Class A up to a higher power output level, although in reality this means very little - the biasing scheme used to keep an amplifier in a particular mode of operation can't make up for for an overall amplifier design that's not of the same calibre and may actually be necessary to compensate for non-linearities in operation. In other words, if you start out with a much better circuit design to begin with, you don't need to over-compensate with higher biasing.
Over the years of owning various classes of amplifiers I have come to realize that excellence in amplifier power supply design is far more important than cross over distortion.
It all depends on the amp design. Typical class AB amp will suffer slightly more distortions with higher bias (overbias) as it is shown here: (chart on page 189)

and here: page 107-108.

The reason for that is transonductance/gain increase (gm-doubling) when both transistors operate (class A) versus one transistor only (class B). Increasing this area of double gain slightly increases distortions. It is perhaps possible to compensate gain with signal level but it would require special design.

Another problem is that class AB amps use deeper than class A (often 10x) negative feedback to correct transistors nonlinearities. You just cannot simply overbias class AB amp and expect miracles. It is not sound quality vs. power dissipation tradeoff - it is much more complicated.
As the posters above wisely point out; there is more to making good sounding amplifiers than just Class of operation.