Some of the changes:

1. The way Men-at-Arms are priced was overhauled, and they were also given a second look stat and pricewise (quite drastically so, in some cases), but that's less important compared to this.

Now, Men-At-Arms only take a token amount of gold to recruit (20% of what they did, and you still pay full maintenance), however hiring regiments also expends prestige, around half of what the gold cost formerly was.

For tribes, Men-at-Arms still cost solely prestige.
Mercenaries still cost solely gold.
Holy Orders still cost solely piety to hire.

Price categories and stats are now standardized. That doesn't mean every culturally-unique unit follows this formula completely, and they are still different from one another, but this is the general outline (for cost/stats, upkeep, terrain modifiers etc still vary depending on unit):

Hiring prices per regiment

(Again, upkeep is another thing entirely)

Cheap Men-at-Arms : 10 gold and 20 prestige.
- Bowmen and Light Infantry, including most culture-unique ones. Generic is 34/30 statpoints (LI has pursuit/screen, bowmen don't), whereas culture-unique troops usually get 6 more points between base damage and toughness than generic variants, i.e 38-40 total.

Cheap+ Men-at-Arms : 15 gold and 30 prestige.
- "Strong" culture-unique Bowmen and Light Infantry (Like Longbowmen, Chu-Ko-Nu, Abudrar). 55 stat total for the former two. High, yes, but they're iconic, and still very killable. Abudrar is weaker at 43, but archers have more upkeep, and more common counters, and don't have pursuit/screen.

Regular Men-at-Arms : 20 gold and 40 prestige.
- Heavy Infantry and Pikemen. 50 base stat total.

Regular+ Men-at-Arms : 25 gold and 50 prestige.
- Most culturally-unique H.Inf and Pikemen. 63-66 stat total.

Costly Men-at-Arms : 30 gold and 50 prestige.
- Light Horsemen - 48 stat total, with great pursuit and screen, and better depending on terrain, as well as the "Strong" Heavy Infantry - Druzhina and Huscarls, who get 68-70 stat total.

Costly+ Men-at-Arms : 35 gold and 70 prestige.
- Most culturally-unique Light Cavalry. They get around 10-11 points over generic variants (58-59 stat total), alongside all the advantages and disadvantages of being cavalry.

Expensive Men-at-Arms : 40 gold and 80 prestige.
- Basic siege engines (Onagers), Horse Archers (65 stat total, with distinct terrain advantages and crippling disadvantages), and "Strong" culturally unique light cavalry, who get +14 over generic (62).

Expensive+ Men-at-Arms : 50 gold and 100 prestige.
- Armored Horsemen (142 stat total but half size), Palatini (Roman Legionaries, a broken Heavy Infantry unit with 90 stat total, which you only get if you unify Rome, and that's a vanilla thing), and Mangonels.

Luxury Men-at-Arms : 60 gold and 120 prestige.
- Trebuchets and culturally unique Heavy Cavalry (Cataphracts and Gendarmes), who get 158/160 stat total respective (again, half unit size, so that's really a 79/80). Also Holy Order Knights, but they are a bit weaker.

Luxury+ Men-at-Arms : 100 gold and 200 prestige.
- Bombards. War Elephants (310 unit total, quarter size so 77-78 - they get a bit lower upkeep than their price, and keep in mind they are not culture, but region-unique.)

Siege engines had their cost and upkeep quite massively increased.

Reasoning (Brief)
Since men-at-arms are basically your household troops, tying their recruitement to prestige in some way makes sense - and it also makes trucebreaking/going negative prestige pretty bad.

But most importantly, this makes AI recruit MaA way better and faster. So instead of waiting 50 years to face the famed french cavalry, you might now need to wait 5, if that! And everyone still pays the upkeep, so its not like they are free.

Furthermore, this also means you can get quite a bunch of retinue up and going from the very start of the game, rather than having to wait for gold to accrue (and you should be able to, since its not like time suddenly started in 1066 and before that realms had no men-at-arms of any sorts).

2. Mercenaries are now slightly more expensive, on average, as are holy orders.

3. Emperors get +2 monthly prestige (because MaA need prestige), Kings get +1, Dukes get +0.5, and Counts get +0.25. This also means Emperors/Kings should field much bigger retinues than Dukes/Counts simply because prestige is limited in quantity. And also because warmongering and winning battles can net you a bunch.

Further Tweaks :

- Duchy capitals get a higher building budget at game start (i.e most should start with 1 more building already built and 1 more upgraded, than other baronies)

- Temporarily added +10 vassal opinion to hard and +20 to very hard difficulties. This is to alleviate the early game spiral of scripted claims and wars, which (due to the AI being better at defending!) causes it to jump from war to war for half of the first century against a baziliion claimants and libertarian vassals. In vanilla they just lose 99% of those, and that's it. Will remove the modifier when I fix civil wars in general, since I'm not a fan of giving AI bonuses and calling it "hard", but for now its a patchwork fix.

- AI is a bit more eager to upgrade its main holdings, compared to the rest of the buildings.

- Warscore from occupation increased slightly. Wars were resolving a bit too slowly on average.

- Added missing localization.

Small fix.

Economy Tweaks...
- Levy upkeep increased by 25% (2->2.5 gold per 1000 raised levies)
- Building income reduced slightly (10% or so, depends on type)

Adopt Feudalism main decision requirements changed
Instead of needing to both follow a reformed religion and to have all tribal era innovations unlocked, you need to either follow a reformed religion and the year must be 1000+, or to have all tribal innovations unlocked, making it significantly easier, though the minimum Fame Level requirement has also been raised to 3 (was 2).

Capture odds rebalanced.

- Capture chance on a succesful siege reduced (50%->30%). This should hopefully make sieging capitals a bit less juicy, and a bit more risky a goal.

- Capture chances for commanders reduced on stackwipes - they were too high previously, leading to stackwiping an army led by an enemy king/emperor being many a time the end of any war, and while that is fine as a possibility, it was far too likely. Furthermore, this lets AI hold onto its good commanders a bit better, because everyone can get stackwiped sometimes.

Numbers to visualize :

Regular battle (Unchanged) :

Was +/- 10% chance multiplied by a factor depending on prowess (67% at 10, 33% at 20, 5% at 29+). I.e 5% chance per lost battle for an average 15 prowess comander, increased if brave, one legged and/or maimed, reduced if craven.

Stackwipe battle (old) : +/- 85% capture chance, affected by prowess / traits / injuries. I.e 15 Prowess commanders had 42.5% odds of being captured (or 45.9% if brave, 47.2% if maimed/one legged, 48.5% both.)

Stackwipe battle (now) : +/- 40% capture chance, affected by prowess / traits / injuries. I.e 15 Prowess commanders have around 20% odds of being captured on a stackwipe (or 28.6% if brave, 33.3% if maimed/one legged, 40% both.)

Do note that characters tend to get injured after several battles in a row - so if you don't capture someone the first time around, chances are you just might still, the next time.

- Tightened building logic for the duchy buildings, since it was starting to get a bit loose.

- Lower level buildings produce a little bit more income, helps the AI kickstart the building (base income of buildings moderately increased, per upgrade level increments left unchanged)
- Reverted an overdone change to the AI spending priority between buildings and men-at-arms, which was causing the AI to build less than desired, but I kinda panicked when I saw fully upgraded level 4 holdings with all buildings level 6-8 everywhere in europe by the 1300's in observer games...