The playtest has some really interesting ideas, but the O5E Sorcerer has 3 major problems: not enough spells known, not enough metamagic options known, and not enough sorcery points (or, a Long Rest is too long to recharge). Almost nothing in this playtest document addresses any of those issues, except splitting metamagic options into Minor and Moderate so that a PC can have 2 meh and 1 good option (formerly 2 good options).
The playtest gives TONS of new ways to SPEND sorcery points, but without enough points to go around as it is, you won't get to use all the fun features.
Conduit: I like that there isn't a penalty converting Sorcery Points from Spell Slots, but Sorcery Points need to recharge on a Short Rest. Also, what is the point of the name?
Manifestation: Weird for this to focus on only elemental damage-dealing spells. Why isn't this more inclusive? The options are bland and unoriginal, and Fire is SUPER good. If keeping this, don't make it a choice - just give them all as options.
Latent Power: Really neat RP options, although they feel expensive without giving SP a short rest recovery.
Metamagic: Similar to Manifestation, IMO just give them all to the player. Or double the amount of MM known. I don't want to play a class if I don't get to actually play the fun aspects of that class, and only having 1 "good" MM until 10th level is just boring.
Arcane Refinement: These are almost okay, but each has something that just makes it awful. Deep - seriously take damage to restore SP? WTF? Patient - LET THEM CAST RITUALS WITH SLOTS! Prodigious - okay, so a Warlock boon four levels later ... meh.
Rapturous Presence: They're okay, I guess... but they feel a little uninspired. And again, more ways to SPEND points, but there just aren't enough.
Evolving Manifestation: These should be additive effects to the originals, and the sorcerer should know them all.
Knacks: Interesting enough, but definitely don't need to be spending SP on these.
One way to solve the “too few spells known” problem is something I call Spontaneous Spellcasting, which I came up with and posted to Reddit. Originally intended for Wild Magic sorcerers, I think it could work for almost all except the new Tasha’s subclasses which know plenty of spells: