I would like to add to my above post by stating that I would rather there be no per-turn mana cost in general, but I thought I'd bring up the idea of having it for enemy players' units as opposed to not being able to charm enemy players' units at all

I see your point Netriak, but I must agree with AM_Shark. It does make gameplay more tactical if there is a limited numer of turns you can control the charmed creature. Especially if you have a 100% chance of charming it. Then you still have to wonder if you should charm it now or wait until a battle is close at hand, but then you might have to use other spells that would prove more useful. You can't go around charming everything and building defenses with everlasting monsters. That said, the other way you can prevent this "everlasting" side is by imposing a cost for charming creatures. I believe that ONE of those two options should be used, but putting both into place would be a little too much. In the end, either inforcing a cost (mana) per turn or a limited time (with possibly an upkeep penalty) would be a good solution in my opinion. I must say I prefer the second option though (with the upkeep penalty).

If you read my previous post, I was the one that proposed the idea that the duration of a charm depended on charm strength and creature/caster level. My point was solely against save-load abuse being something that Stardock should balance about.

Concerning infinite armies:

That has nothing to do with the charm spell. It is simply one way to acquire units. The difference is that instead of gold, you pay mana as initial cost. Instead of being limited by population, it is limited by availability of wandering monsters. But it is fundamentally just another way of acquiring units. I agree there should be an upkeep cost, but all units should have one. The more fantastical creatures may very well have a mana (As in the resource, generated by shards. Not unit based.) cost as upkeep instead of gold.

Having a low upkeep for neutral critters to be charmed might not be such a great idea. In any situation where a small kingdom could charm a bunch of them to hold off a larger kingdom, the larger kingdom could charm them quicker/earlier and have an even larger army.

It would also be cool if we had some brainsucker parasites. They could attack a unit in combat, suck their brains right out, and crawl into the skull themselves. No upkeep for the unit, just for the parasite. The unit gets large combat bounses since it dosen't care about living anymore and ignores daamge. The catch though is that the unit falls apart in some set amount of time. After all, the parasite isn't taking care of the body, just using it as a combat drone. And, we should be able to infect our own forces. Line up a group of peasents, pop a parasite into each. They go from being level 1 commoners to berserker golems with the parasite's vast experince guiding it. Would be hard on the local population but handy if you are in a pinch and need some troops quick.

Or we could have veteran parasites who could infiltrate generals/princes/leaders of other armys and lead them into ruin.

I dislike 'all or nothing' binary states, as they encourage save/reload play, and are generally either really broken, or really useless

Ditch the % chance of success (save/reload abuse, or randomly lucky power gain), and instead simply tie the charm power to its potential targets

A level 1 charms level 1 foes, a level 5 charms level 5 foes, etc

Play around with that, and if it works fine, leave it, if not, add an interesting downside to make casting the spell an interesting decision, instead of either a roll of the dice or an overpowered ability

eg, once charm breaks, the monster rampages against you with increased strength, or you suffer increased aggression from its relatives (mommy dragon don't like you), or you trigger random events as a consequence (troll rebellions due to your abuse of their kind!), or your foes empire gains temporary morale bonuses against you (he made me fight my brother!) etcetc