Hi there. Had to have a quick torpor tantrum, but I see that little has changed here, so I'll continue my discussion with looking at swords in more detail. This discussion predicates on my topics about status procs and family damage and weapon switching and finishers, so please at least skim those if you have not already. For damage numbers, I will make reference to the wiki's recorded numbers for the end of S6; keep in mind these don't account for things like damage bonuses favoring split damage — I'm using them for quick reference because they're the best data I have on-hand.
Normal Swords
We must begin with Brandishes, and their high king Acheron. Brandishes are notoriously strong, but quite frankly, I would leave them be — with the exception of Acheron, whose damage should be normalized, and then for its charge either given more damage specifically or a status, to make it match with the way other status variants work: since Brandish normals don't have a status, Acheron normals doing more damage is nonsense.
Broadly speaking, swords which are not Brandishes would be in a much better position simply by having more damage. Most swords are fundamentally okay and are only outclassed because Brandishes are Just Better™.
Leviathan Blade really just needs to be a bit stronger. Its largely a fine weapon, but its charge is riskier and requires more setup to use well, but hardly pays off. Combuster Charge vs eg Slime is 531 + 128*5 = 1221 under ideal conditions, vs Levi's 386*3 = 1158. Getting three hits with a Levi charge is difficult, requiring good positioning from the user and the enemy to be frozen or otherwise pinned so that the knockback doesn't launch them out of the later hits; not to mention, Levi requires the user to get close to do any damage at all, Combuster et al. have status procs, damage which scales better due to a quirk in the defense formula, and significant bonus damage vs elemental weakness in a game where it is generally easy to change out weapons based on upcoming enemy resistances.
There's a lot of factors and wiki numbers are a bit simplified, but the point is Leviathan Blade is a risker weapon but with less payout. Its normals are the same or weaker most of the time, and its charge is fun but also not competitive. I would advocate: 1) bump up its base damage so its normals are a sort of midpoint between Combuster's Effective and Neutral normals*, 2) buff swing 3 to be DPS positive and reliable flinch on hit, and lengthen the hitboxes for swing 2 and especially swing 3 (this applies to all swords w/ the same set of normals, so Brandish-type weapons and Wrench Wand (lol) too) because they struggle to connect reliably without careful timing and midstepping, which is a bit ridiculous; midstepping should be tech for advanced play, not necessary to make the weapon work at a basic level.
(*this might be a general rule for Normal damage variants, but here the specific concern is Levi as the melee knight sword being totally beaten by the safe wizard sword)
Ok, whew. How many more of these are there? 16? OK. Going forward, my general rule: if a Brandish would be an alternative that is safer and more effective, the weapon needs changes. Most swords are riskier and changing that compromises their identity, so most swords just need more damage. There doesn't have to be a mathematically perfect balance (since that's impossible to do in a fun way), just as long as there is not an obvious superior alternative.
Cold Iron I would advocate for being an Normal/Elemental +Freeze Levi. As common as split ele (with status) is and much as I like pure normal, this maintains its +vs Undead in a way that actually performs well and isn't gimped by the game's mechanics. This also would allow Sweet Dreams to maintain a meaningful difference from Celestial Saber (since both have stun. Or SDreams could be Stun+Freeze, it's prizebox semiclone of a bad weapon what do I care.) Again, finisher should be DPS positive over not using it and safe on hit vs most enemies (that don't resist it). CIV and other Levi variants should appreciate a comparable damage increase if Levi gets one, except C. Saber I guess since it has Levi Damage and a Status to begin with. C. Saber also apparently doesn't flinch on hit 3 (I can't check I'm poor) so either its finisher should be tuned to be worth using or it should just...not do that.
Fang of Vog is weird. It's mostly a Calibur, but with split elemental and fire, lowered base damage to compensate for the status, slower swings, and slightly longer reach; its charge: has an easy-to-land additional hit, reliably procs powerful fire, and sets you on fire (apparently the self-fire hurts more if you have damage bonus??). It's pretty much a better Levi. I'd probably still scale it proportionately to Levi, though, since its drawbacks make it feel reasonable compared to the normal Levi. FoV does seem to proc reasonably often with normals, but I haven't tested to see how it performs vs eg switching Magma Driver.
Fast Swords
Next up is the Flourish (and Barb). These weapons have...good normals. They're a bit weaker than Combuster etc but also a bit faster. They do need a better hitbox on swing 2 since it rarely connects without midstepping. The only variation in change I'd consider is making the both the second and third swings stronger, since the second swing has less coverage and more commitment than the first (this is why spamming the first swing w/ shield cancelling is common for Flourish).
Flourish charge is...bad. It's not particularly safe since it means diving straight into enemies, and it again does less damage than landing every hit of a Combuster charge. I think the charge itself is fine as long as, once again, it has worthwhile payoff. Barb charge is just bad, with its included backward step requiring the user to get deceptively close to the enemy to use and its damage output is just plain poor. You'd be lucky to get 1000 damage out of it even against a piecing-weak enemy (though admittedly it's also hard to tell because all the damage numbers visually overlap). It's a Brandish charge that's harder to maximize, riskier to use, and does a lot less damage. And of course, Rigadoon and Flamberge need better status chance because they are unreasonably terrible at status application.
Cutters are in a bit of an odd spot. The charge attack is decent (still less safe and less damage than Brandish!), and choose between WHB's reliable flinching and DVS' poison chance, though strangely the final hit can be cancelled. Cutter charge may seem like a flurry of normals, but 4/5 hits do 50% more damage, so it's decent illustration of choosing between safety (normals) and efficacy (charge). The weapons probably would appreciate a bit more base damage, though, since the strings themselves are safe through their speed and not from flinching — landing the finisher is a lot of commitment and usually does not flinch. I've heard the suggestion that Cutter normals should allow free movement, which I have mixed feelings about, but would certainly fit with the high-speed and low-risk the attacks seem intended to present.
For DVS, ideally its normals would proc poison more often, since the existing proc chance gives it massive droughts in usefulness. Make the normals have weaker poison if it feels too free.
WHB, on the other hand, is A Whole Thing. First: why are its normals slower than DVS? As far as I know, it does not possess superior damage or range. It's just slow. Second, abolish the Beasts High property and make its damage Normal/Piercing. Finally, what's with the flinching? There are a few weapons with anomalous flinch behaviour but for WHB in particular the change in safety between its charge and DVS' is night and day. If flinching more reliably with the same moveset and damage is Its Thing, it really deserves a tooltip mentioning that. If it, in general, can flinch a lot more reliably then its lower speed makes a bit more sense. If the flinching is just a charge attack behaviour, then its normals do not deserve the speed loss. I'm not sure I could say that WHB should do more damage though, since the charge flinching constantly is easily equivalent to having a status proc in terms of benefit.
As a final note, Cutters are one of the weapon types that make people desire a way to attack repeatedly without mashing. Maybe just add a separate binding for that? I don't see a reason to make 'autofire' a property of particular weapons, nor do I think it would be good for them to lose charge attacks to get a QoL feature.
The other fast sword, Turbillion, is actually very strong, if switched. Its normals are quite laggy but cancel very early, except for swing 3, which as a finisher doesn't cancel and so is very slow — it desperately needs that increased combo finisher damage. Turbillion is a bit of an anomaly, though, with better damage than a level 10 Winmillion out of the gate (Winmillion is really bad for no reason) and a better attack speed (Winmillion is slower than Turb/Arc Razor/Spur; Winmillion is really bad for no reason).
The pain point of the Spur family (other than Winmillion, which — say it with me — is really bad for no reason) is its charge attack. Not only is its windup incredibly long, but the sword swing has a pitiful hitbox, and the bullet seems iffy for hit detection as well. It's essentially a worse Brandish charge: riskier to use, harder to land, less damage output, and shielding too early can cancel the active part of the attack. This is a charge attack I would advocate for replacing entirely, especially since Turb normals are already about poking with shots or getting very close with its big lunges to wail on the foe. Maybe the projectile could fly faster and enemies towards the user, especially with Turb charge having the tornado effect — pulling foes in would chain nicely into the aforementioned wailing with normals.
In general, for Spurs, I would change the charge attack to be more unique, or at least fix the hitboxes and the animation cancelling too early. I'd put at least some version of the bullet on Spur and Arc Razor normals (their tooltips even suggest they should, with text like flings sharp bolts of energy outward when swung) — it could gain increased range with higher star level, or something. And, fix Winmillion. Please. There's no reason for it to swing as slowly or perform as poorly as it does.