You have Kraxis and Tarth twice

Please give us more info. How many AIs per game? Which ones? 

Means Tarth and Kraxis won twice.

Frogboy, can you also add at what turn the game ended?

No surprise the two magical ladies aren't up there indeed. Maybe the obnoxious -25% efficiency could be reduced to something more sensible like 10%?

At least the Kingdom v. Empire dynamic seems balanced from this limited testing. I would also like to watch an AI war, that'd be kinda cool. 

How do you start with cheat mode enabled... is it an .ini switch?

Yeah, it never ceases to amaze my how unbalanced the weakness picks are.  You go from "Absolutely harmless with a modicum of care" (Clumsy) to "Seriously nasty" (Cruel) to "Damning" (Inefficient)

Same with professions: How can you possibly think that "Bandit lord" and "Hunter" are in any way comparable to "General" and "Armorer"? I'm not even talking about professions which can decide a game when properly leveraged, like "Beastlord", which can get you a Black Widow on turn one, and a great wolf when you next get the mana.

Of course, depending on the difficulty level, weaknesses and advantages shift in importance.  Karavox does much worse on Ridiculous+ and Procipinee much better.  From Derek's outcomes, I guess the AI were challenging.

I have been playing standard sovereigns lately, and they are all very poorly designed for their purpose... Which I guess is OK from a lore prospective.

Yea, wealthy is pretty much nothing on higher difficulties.

I don't know how much the AI changed since 983, but my game ended with a big knock-down drag-out between me (Lady Porcupine, apparently) and Relias. Relias had been doing very well and conquered Yithril, Gilden, much of Tarth and Magnar (before they surrendered), and so out-powered me about 450 to 150 (or maybe more, don't remember exactly). I think the AI leveraged henchmen quite well as long as it was winning: they kept leveling and got better equipment, etc. Once I pulled off several major tactical victories (through extensive magic use) they seriously underperformed without good enough supporting troops, but I'm not sure he ever stopped building them. By the time my armies caught up to his they were a liability, either by having too many wounds to be useful and taking up army slots, or by being outright negative with dysentery or whatever that wound is that gives -x% to army hp. In fact, when I started rolling up his cities he lost any chance he had as all his henchmen would respawn in cities but be immobilized, and then when I attacked his troops were at a total -50% hp thanks to multiple cases of cholera (or whatever). (Not to mention Relias had a base hp of 3 thanks to the useless familiars he kept summoning, but I take it that has been fixed.)

[Re: Quest Maps: does the AI even do quests?]

Quick fix: don't let malaria (or whatever it is) stack.

AI fix: teach the AI when a hero or henchman becomes a liability and have it put him/her out to pasture.

Mechanics fix: have heroes (or at least henchmen) die for good at some point. Alternatively, perhaps tone down wounds a bit (and/or put in a healing mechanism) but cause the hero to drop an inventory item upon defeat. (So the setback is serious but temporary.)

Aside: I think "sidekick" is a better term than "henchman" because henchman implies a faceless goon whereas sidekicks traditionally have some personality, but less of one than a hero.

Inefficient is not that bad.  I take it all the time.  The penalty is only applied when the sovereign is actually in the city.  My sovereign spends 99% of my time in the wilderness, thus I get no penalty for 99% of the game.

I tested it out in a quick game in .992.  I had a building requiring 11 more turns to finish with the sovereign in the city.  It dropped to 8 when the sovereign left the city. 

Oh  I assumed a 6 AI free-for-all. Some more details on the AI Wars would be nice

efficient would be a better trait functionality. It would cause a real tatical question of to leave the sovereign in city or adventure,

Inefficient I thought applied always...Who keeps their sovereign in town for long periods of time? I think this is kind of bugged because it turns out not to be a real weakness at all. I would suggest implementing it always, and have the penalty turned lower to 25%.