Greatsword Warrior build: the self-healing auto-battle king

✍️ Alice・愛麗沙

Free-to-play and mid-budget sheets side by side are one click away in the loadout builder; the raw per-slot annotations sit on the class page.

The self-heal is written into the passives, not into player folklore

Two of this class’s passives interlock. One says: every time you spend a certain amount of HP, attack power gains a stack. The other says: when an attack lands, restore health scaled by the number of stacks the first passive has built (official skill text). Translated out of the tooltip: the class treats its own health bar as a resource, spends it for damage, and earns it back by connecting. Bolt on the third passive — defence rises while a skill is winding up, and the wind-up cannot be broken — and you have a character that gets healthier the longer it stands there trading blows. It is, frankly, unreasonable.

The Korean-community verdict on its idle performance is blunt: the best self-heal in the game, ranked first for long unattended sustain. Taiwan board reports are more concrete still: greatsword on full auto never takes a step back, and a 19,000-combat-power alt cleared hard small-scale Glas on auto (player-reported, corroborated across threads). The other half of the auto-battle duopoly is the Frost Mage, which wins by keeping its own conditional runes fed. The greatsword wins on one thing. It does not die.

The honest counterweight belongs in the same paragraph. Korean players at launch complained the class was worse in low-level field content — the animations are slow enough that mobs die to somebody else before the swing lands (player-reported, S0-era). The greatsword’s idle advantage only starts paying once the content hits back hard enough to matter.

The sheet, slot by slot

This table is generated from the site database at build time. Full loadout: Vanguard

The same loadout sheet read across all three budget tiers, slot by slot. A cell marked "no source found" is a real gap in the research, not a missing entry.Full loadout: Vanguard

Three judgements the table cannot make on its own.

The weapon slot is where this article takes a side. The Korean graduation sheet puts Thorn Vines at the top — damage taken and damage dealt both +25% — while noting in its own margin that the damage gap against Thousand Swords is small. The Taiwan hundred-GP write-up orders it differently: Thousand Swords first, with Thorn Vines only third and Duel+ last (player-reported). We side with Taiwan, and the reason is the class’s job description. You level a greatsword mainly to leave it running — and unattended, that +25% damage taken is being paid directly out of the self-heal loop that is the whole point of the class. Thorn Vines earns its slot when you are hand-driving a dummy for peak numbers. Not before.

The top and bottom slots make this the only class running Torch+ and Icebreaking+ together. Those two runes require attacking a burning and a frozen enemy respectively — and the greatsword applies neither status itself. It relies entirely on a pet (the Taiwan write-up flags the same dependency). Without the matching pet, the Korean sheet spells out the fallback: drop to Stability+ (attack +9% above 75% HP — a condition the self-heal keeps live almost permanently) or Enlightenment. This is the textbook case of copying half a sheet and ending up worse off: a legendary whose condition never lights is beaten by an elite whose condition always does.

The necklace is the one hard-coded slot in the entire dataset. The Korean sheet specifically instructs that Rotation+ must be engraved on the necklace — the only accessory rune in this file with its slot pinned. It gives no reason. Follow it anyway; pins like this usually trace back to the cost of re-engraving later.

Momentum: the engine the whole rotation runs on

The feel of this class changes at one rune, on one ring: Momentum+. It converts skill 4, Cleave, into a charge skill, and the rule is the interesting part — using Cleave banks a momentum effect scaled to its charge level, and any other charge skill then spends that momentum to jump straight up the charge ladder. The Taiwan write-up compresses the loop into one line: Cleave, momentum buff, then an instant Wind Cleave with no wind-up.

So the greatsword’s rotation is not a memorised key sequence. It is inventory management on a momentum stockpile: skill 4 stocks it, skill 5 spends it, and skill 1 — dramatically faster once modded with Rotation+ — fills the gaps. The Korean S1-era baselines were a 4-2-1-5 opener into a 1-4-3-5 sustain, topping up with 1-4-5 on a cadence.

The third accessory is a choice about how you spend the stockpile. Counterattack+ turns skill 2, Rising Smash, into a charge skill with doubled super armour, and taking a hit mid-charge snaps you to full charge — combos stop being interruptible, which is the stable line. Resolve+ makes skill 5’s crit rate scale with charge level and doubles it against low-health targets — the crit line. The Korean sheet’s advice for under-geared players is Counterattack+, and it is right: below spec, not getting your combo broken is worth more than crit expectation.

Update 9/14 (Taiwan server): Taiwan’s 9/9 patch notes (thread 3541097) touched Greatsword in three places. Counterattack rune — taking a hit during use now grants 1 Battle Rhythm regardless of stance. Severance rune reworked — it no longer applies the Wound damage-over-time; instead max uses become 2, with longer cooldown and higher damage, so any old setup using Severance as a bleed engine is dead. Retribution — fixed to persist after combat ends, stack cap now 100. The Thousand Swords / Momentum+ core is unaffected.

One thing Taiwan players raise that the sheet does not cover: the greatsword has two grips, and skills 2 and 3 change behaviour in the reverse grip (there is official skill text for it, and the Taiwan write-up maps the timing). Idle players can ignore this entirely; manual players should read the raw annotations on the class page.

Mods and level-30 enhancement: 1·4·5 is default, 2·4·5 is emblem-gated

The Korean sheet’s default is 1·4·5, and it names exactly one exception: switch to 2·4·5 once you have the Wisdom emblem (all skill cooldowns −18%), because only then does the Rising Smash axis turn over fast enough to be worth running. A 1·2·5 layout was briefly the community standard at launch, but that write-up said in its own text that there was no settled answer, so treat it as history. And the drift toward 4·5·6 belongs to Korea’s later seasons — it has nothing to do with Taiwan. Do not copy it.

Level-30 enhancement order is 1 → 4 → 5. The Korean reasoning is that skill 1 appears three times in a single cycle and therefore carries the largest share of damage. The Taiwan hundred-GP write-up ranks it 4 > 1 = 2 = 6 > 3 (player-reported). Notice that both put 1 and 4 in the top two and disagree only about the order — so our reading is: follow the Korean sheet and finish 1 first, since it is also a mod target, but if you run the Counterattack line and lean on skill 2, the Taiwan ordering is entirely defensible. Skill 3? Here the two sources agree completely. Do not invest.

The free-to-play caveat, as always, arrives early: elite-grade runes carry no all-skills+1 line, so level-30 enhancement is out of reach without spending in the first place. For a free account the mod question collapses to “which skill does each of my three accessories rewrite”.

Carving, cards and seals: follow the smash

Gem tags run smash · support · support (stated for this class in the Korean data; the Taiwan write-up’s “prioritise smash lines” points the same way, player-reported). Carving priority: smash-damage and support-damage lines far ahead of any cooldown line — the cooldown that the momentum loop wants comes after the damage lines are full, not before. Prism sub-stats go crit > smash-enhancement > multi-hit-enhancement, with smash stacked all the way to 6,000–7,000.

Upgrade cards: the attack card is always first. The honest disclosure is that our Korean data records no greatsword-specific card guidance at all — the “STR plus WILL plus SPEED is correct” claim exists, but from a single low-confidence source, and we are not adopting it on that basis. So run the general damage-dealer rule: main stat STR, and re-roll the WILL slot to whatever your build actually converts. (Greatsword’s registered sub-stat is WILL, which is precisely why that single source is plausible and precisely why we still want a second one before printing it as advice.) Once maxed, pay the 3,000 gold to swap out defence, HP and movement-speed cards one by one — the three perennial damage-dealer traps. Star seals go on the weapon first (+300 attack), then the emblem (+10%): a multiplier needs something to multiply.

Group content: S0’s god, and then what

In Taiwan’s current raid environment the greatsword has a very physical advantage. Glas Ghaibhleann’s pushback and terrain-destruction mechanics send most classes travelling back to their spot; the greatsword simply stays planted and keeps hitting (Korean community consensus at the S0 timepoint). Its self-healing is good enough that Korean players reported soloing an eight-player raid.

The same coin’s other face shows up in the abyss. The Mars abyss does not restrict elixirs — with potions available, other classes hold their ground perfectly well, the self-heal premium collapses, and Korea’s abyss rating for the class fell from an early S to B. One exclusive contribution survives that demotion: in the Hall of Ruin’s large-ogre room, sending in the class that can stand and trade to hold aggro first cuts the whole party’s clear time substantially (Korean).

The same Korean sheet also records a standing flaw that the Taiwan discussion rarely mentions: in boss rooms, a five-second internal cooldown on the wound effect leaves skill 5 sealed for a fifteen-second stretch. We have not reproduced that timing ourselves, so treat it as a Korean-sourced structural note rather than a measured Taiwan number — but if your boss damage feels like it stalls in a repeating pattern, this is the first suspect.

Lookahead: the developers conceded the weakness themselves

At Korea’s class-balance stream on 2026-08-12, a developer said something about the greatsword that developers rarely say out loud: strong in solo play, but its damage contribution in a party tends to be low, and they are looking at it (Korean, community-relayed). Translation: the official position now matches the community consensus — king of idle and solo, second division in group content.

Korea’s Season 2 answers it by giving the class a new Berserker-style burst axis, with the mainstream mod layout shifting toward 4·5·6 and two named styles emerging — a high-spec burst build and a stability build. All of that is Korea-only lookahead. None of it exists in Taiwan, the runes it is built on are not in the Taiwan client, and Taiwanese names for any of it will be whatever Nexon Taiwan ships. The practical instruction for a Taiwan player is short: play the self-heal king with a clear conscience, and do not stockpile resources now for Season 2 runes you cannot yet see.

What we do not know

Our card data for this class is a genuine hole, not a summary — it is recorded as not found, and the plausible-sounding stat triplet has one weak source behind it. The boss-room skill-5 seal is Korean-sourced and unreproduced by us. And the tier movements above are opinion data from community sheets, dated and attributed but not measurements; the abyss S-to-B fall in particular is one community’s re-rating, not an official change.

Our verdict

The Greatsword Warrior is Taiwan S0’s best class when nobody is watching. The self-heal loop and super armour give it the highest unattended floor in the game, and its gear bar is low — outside two core accessories it wears the generic armour set. Our recommendation: if you want one character that runs dailies on its own, farms barter goods as an alt, and occasionally gets driven by hand into a group, the greatsword is the first pick. If you want to top the damage table in an eight-player raid, it is currently not that class — and the developers have now said so themselves.

A claim you can overturn: at equal combat power and eight hours unattended, the Greatsword Warrior’s survival rate in Taiwan’s current field content and small-scale Glas beats all seventeen other classes. If a Taiwan test — same content, same duration, death counts side by side — shows another class holds up better unattended, the “self-heal king” framing on this page gets rewritten. The cross-class power ranking is at the tier list.