It is a homebrew campaign. We startet at lvl. 3 with a one-shot. At that level the Mystic looked fine, but the higher it gets, the more powerful he is, against single targets and multiple targets. He is an Order of the Avatar-Mystic and uses often the Mastery of Fire Discipline, which does a lot of damage. Detonation at Mystic Lvl. 5 allows for like 5 Fireball-Exquivalents.
Now at Lvl. 7 the character took psionic restoration and can heal better than any cleric. He can casts like 7 times reveviviy before he runs out of psi points.
The Mystic is Mary Sue.
With Mantle of Joy it is great in a social role, as good as a rogue. With Master of Fire it is as good as a wizard casting damage spells, with psionic restoration it can heal better than a cleric.
All put together in one Character. It can do too much to good.
I don't see why that is a problem. It's not as if the same character could do all those things, you still had to choose. So far as I can see it makes no difference if you make that choice at "choose a class" or "choose a subclass" level. Mechanically, the outcome is the same, and it cuts down on repetition. You only have to explain psi points once rather than copy-paste it into several classes.
In 5e, subclasses have a certain amount of mechanical heft to them. This heft is limited.
When they melded them together in the Mystic, you ended up with combinatorial explosion of abilities. You got a chassis that (in theory) supported a full featured melee character and a ranged spellcaster controller. What more, you could basically be both if you wanted.
And then there is the core class features. They must be suitable for all 3 flavors (tanky, skirmishy and castery) as written; if you split it, the core class features also split.
You'll notice that I hijacked one of the subclasses for Soul Knife and Immortal, then layered a subclass on top of each. That gives a lot more design room to build a skirmisher or tank, and we don't have to either (a) put a full "kit" into one discipline, or (b) make disciplines useless unless you combine the right set of them. Both of those problems are in evidence in the Mystic.
Combinatorial explosion is also a big problem as written. Notice above, I took a bunch of abilities which (alone) are at least somewhat reasonable, stacked them on top of each other, and got someone who can do the damage budget of 3 top-level spells in one action.
So we end up with it being challenging to construct a character, challenging for designers to know if the class is completely broken OP or not, unclear what kind of archetype the subclasses actually represent, subclasses and classes that are mechanical bundles of abilities more than coherent story seeds.
As a concrete example, the complex mastery concentration mechanic looks great for a psion who is some kind of cerebral meta-psionic character. For a soul-sucking psionic vampire, a psionic body-melding perfectionist, empath who emotionally boosts their team, it is looks sort of strange.