Class balance

How the 11 classes were evened out in PvM and PvP, at levels 400 and 1000 — with the simulation numbers and what changed.

No class here was designed to be the strongest. The numbers below come from a simulation that uses the real server logic — the same code for combat, skills, Master Level and items — with all 11 classes under the same conditions. The goals: every class's damage per second within 10% of the average against the same monster, and a 1v1 fight between similar characters lasting about 10 seconds, with no matchup above 65% wins. In the game, the ⚖ Balance menu shows the same numbers.

How we measured#

• Characters: all 11 classes at their 2nd evolution, level 400 (no Master Level) and level 1000 (400 + 600 Master Level, with the whole tree built for damage). Points: the class's own + 5 per level, no Reset.
• The 4 classes from the newer seasons (Grow Lancer, Rune Wizard, Slayer and Gun Crusher) joined PvP on 10/12/2026, with the same characters, sets and goals: their numbers were tuned without touching the matchups among the other 7. In PvM their damage is tuned by real hunting (a spot with the MU Helper) and by bosses, not by the single-target chart — which is the reference only for the 7 classic classes.
• Damage builds: DK strength/vitality, DW energy, Elf agility, MG strength/energy, DL strength/command, Summoner energy, RF strength/vitality, Grow Lancer strength/agility, Rune Wizard energy, Slayer agility/strength, Gun Crusher energy — gear requirements first, the rest in the build's proportion.
• Gear of the same tier: the class's 380 set and 380 weapon, all +15 with luck, +28 option and full excellent; 2nd wing (level 400) or 3rd wing (level 1000) +15. The Dark Lord carries Dark Raven and Dark Horse (the class kit). No gems, VIP, pets, potions or seals.
• PvM: the same monster for everyone, standing still, for 60 s — single target and area (5 monsters together). On each hit the character uses the skill that deals the most damage and is available (cooldown, mana and AG count); the 5 strongest skills are tried first and the DK tries every combo. The rotation with the most damage is kept. While hunting and in PvP, a large mana potion is used when mana drops below 35% (like the MU Helper, and like a player does in a fight); no life or SD potions.
• PvP: 1v1 side by side, both hitting from the first blow, until someone dies. Each class pair fights dozens of times.
• Defense against players: a hit deals hit² ÷ (hit + target defense) — a hit equal to the defense deals half, a hit far above it loses almost nothing, and no hit drops to zero. In the original MU it was hit minus defense, and whoever hit close to the target's defense felt any gear difference far too much. Against monsters it is still hit minus defense.
• Reflection against players: it returns a percentage of the damage the hit dealt, the same for every class, as in the original MU (since balance 5.7). Before, it came back at the PvP scale of whoever reflected, and classes with high raw hits (Dark Knight, Slayer, Grow Lancer) took back 2 to 7 times the nominal reflection: the same fight flipped depending on whether the set had the reflect option.
• Rage Fighter: the flat skill bonus (vitality/10 on Chain Driver, energy/10 on Dragon Lore, agility/8 + energy/10 on Dark Side, +100 from Dragon Slayer against players) comes in before the target's defense, like any other class's damage. In the original MU it came after defense, and part of the RF's hit did not feel any gear at all, neither its own nor the target's.
• Three conditions: no buffs; with the class's own buff (recast when it ends, spending the hit); and in a party, with the other classes' buffs too. The goals apply to all three.
• Other gear: the same test runs with everything +13 and with 3-option excellent (weapon with mana on kill, speed and damage per level; armor with life, damage reduction and defense rate), without buffs and with the class's own buff. The targets hold in all three sets: damage per second on monsters within 15% of the average (10% on the full set), average fight of 8 to 12 s and no matchup above 65% win rate.
• Speed: every character hits at the exact pace of their own attack speed (or magic speed, for magic skills). Every point of speed counts: 5% more speed means 5% more hits, with no steps. The test builds do not move stat points around to chase speed.

Damage per second against monsters#

Against Iron Knight (142), with each class's best rotation. The "before" column is the game before this balance, without buffs.

Level 400

Level 1000 (ML 600)

PvP 1v1#

Each cell: the row class against the column class — average time until someone dies and how often the row wins. No buffs.

Level 400

Level 1000 (ML 600)

Strongest skills of each class#

Damage per second of each skill used alone against the test monster, at level 1000, relative to the class's strongest.

Adjustments in effect#

How much each class deals to monsters and to players, and how much it takes from other players, compared with the original game. Between 400 and 1000 the adjustment shifts gradually with Master Level.

With other gear#

Each class's difference from the average in damage on monsters and its average PvP win rate (against the other 6), no buffs, with other sets — so the balance does not depend on one specific set.

PvP with Resonance#

Against monsters gem Resonance counts in full (up to +500% damage and defense). Against players it is worth less, on both sides: with k = 1 + the Resonance bonus (1 without gems, 6 with everything at grade 10), the attacker's hit is multiplied by k^0.48 and the damage that reaches the defender is divided by k^0.73. Resonance's extra defense does not count against players nor raise SD, and against players the gem effects that look at life (Ruby, Kalima, Crywolf, Noria, Pearl) look at life together with SD — Blood Ruby's life steal restores life and SD. Result: between two characters with the same Resonance the fight gets a little longer as the grade goes up (~10 s without gems, ~15 s with everything at grade 10), and whoever has more Resonance wins comfortably, without killing in a single hit. The table uses the same 4-gem kit on both sides (Heart of Balgass, Shadow of Kalima, Blood Ruby and Tarkan Amber — each gem's effect is worth half against players), full set, no buffs.

• Grade 10 against grade 5: grade 10 wins 100% of fights at level 400 (in 7.4 s) and 100% at 1000 (in 7.2 s).
• Grade 10 against grade 0: grade 10 wins 100% of fights at level 400 (in 5.6 s) and 100% at 1000 (in 5.4 s).
• Grade 7 against grade 5: grade 7 wins 97% of fights at level 400 (in 10.5 s) and 96% at 1000 (in 10.4 s).

Player reports and automatic balancing#

• Report: in the game's ⚖ Balance menu, Report tab, tell us what happens (against monsters or players, your class and against whom, up to 600 characters). Your character is attached automatically: level, Master Level, stats, life, SD, defense, equipped items and active buffs. One report per account every 6 hours.
• Analysis: an AI reads the report and picks the test (level 400 or 1000, no buffs, own buff or party, which matchup). The simulation with the real server logic decides, not the AI: if the class is more than 10% off the average in PvM, or the matchup goes over 65% wins (or the fight leaves 6–15 s), it is a system issue. Within target, the report comes back with a tip (build, gear, skills). The answer shows up in the Report tab and in the chat.
• Automatic balancing: at most once every 3 days, at night (3–6 AM Brasília time), the system runs the full simulation. If there is a report confirmed as a system issue or any target off, it computes an adjustment where each number changes by at most 5%, simulates again and only applies it if things get better without pushing any other target off. It takes effect right away, without maintenance, with a chat notice.
• Transparency: every automatic round goes into the ⚖ Balance panel history with what changed, why, and the numbers of the reports behind it. The administration can pause the cycle or go back to the previous version.

What changed#

All 11 classes in PvP and the same reflection for everyone · v5.7 · 10/12/2026

• Grow Lancer, Rune Wizard, Slayer and Gun Crusher joined the PvP balance, with the same characters, sets and goals as the other 7: a fight of about 10 seconds and no matchup above 65% wins. Before this the Grow Lancer won 98 to 100% against every class and the Slayer lost almost everything.
• Reflection against players: it returns a percentage of the damage the hit dealt, the same for every class, as in the original MU. Before, it came back at the PvP scale of whoever reflected, and classes with high raw hits (Dark Knight, Slayer, Grow Lancer) took back 2 to 7 times the nominal reflection — the same fight flipped depending on whether the set had the reflect option.
• The 4 new classes' own buffs (Wrath, Obsidian and Circle Shield; Burst and Haste; Detection and Demolish; Fixed Fire) now count as "class buff" in PvP, like the others'. The 7 classic classes barely changed: the matchups among them were retuned together, by a few percent.

Even on spots and bosses: adjustment by the monster's defense · v5.6 · 10/12/2026

• Against high-defense monsters, whoever hits many weak blows (Grow Lancer, Slayer, Dark Knight, Magic Gladiator) barely scratched: on Balgass it was 2 thousand per minute against 37 thousand for the Rune Wizard. Now the hit has an adjustment by the monster's DEFENSE — the same for bosses and spots, because they are all monsters — and on the tested bosses (Balgass, Nightmare, Erohim) the 11 classes ended close to the same average. Below 660 defense nothing changes, and the adjustment fades with Master Level (level 1000 already had its own balance).
• The area hunting adjustment (5.5) was redone together, so spots stay balanced with the new adjustment: the average damage per minute while hunting stays the same.

Area hunting balanced between classes and bosses immune to poison · v5.5 · 10/12/2026

• Hunting on a real spot, the area magic classes took up to 3 times more life per minute than the single-target ones (Elf, Grow Lancer, Slayer). Now the hit on the EXTRA monsters of an area has a per-class adjustment: less for Dark Wizard, Magic Gladiator, Rune Wizard and Summoner, more for Dark Knight, Elf, Dark Lord, Rage Fighter, Grow Lancer and Slayer — all close to the same average. The main target, bosses and PvP don't change.
• Bosses no longer take the poison that removes % of life (Poison, Decay): it took 3% of the boss's life every 2 seconds and melted any boss with Dark Wizard, Magic Gladiator and Rune Wizard. On regular monsters the poison stays the same.

New-season skills working and the MG renewal · v5.4 · 10/10/2026

• The high-numbered new-season skills (Chaos Blade, Havok Spear and the MG charges, Raining Arrow, Focus Shot and the Elf's new buffs, the Dark Lord's Crown Force and Battle Glory, the Ruud Shop ones) did not fire: the hit turned into another skill. Now they fire, and the balance counts them. To keep the average damage against monsters the same, the Elf's PvM multiplier dropped 12% and the MG's 14% at level 400 (7% and 5% at 1000); the other classes changed 1 to 3%. PvP stayed about the same.
• Magic Gladiator renewal (Season 16): two one-handed swords (identical ones deal 65% each) or one two-handed sword; or two staffs, with 30% of the left one's wizardry. Spiral Charge and Fire Blood need a two-handed sword, Crusher Charge and Ice Blood a one-handed one, Elemental Charge a staff. Fire Slash costs 12 AG.

Season 16 Master tree · v5.3 · 10/9/2026

• Season 16 Master tree: the new classes got the enhancement and mastery of their skills, 3rd wing and weapon; the original ones got the enhancement of the Season 12 skills and elemental defense. Where Season 16 only renumbered a node the class already had, it is not added again (it would count twice). This changed the level 1000 tree a little, and the adjustment corrected each class's damage and defense by 1 to 5% to keep the goals (Rage Fighter was the most affected: the new Phoenix Shot nodes took points away from PvP).

Area around the target · v5.2 · 10/2/2026

• Area around the target: Death Stab (DK), Fire Burst (DL), Chain Lightning (Summoner), Dragon Lore and Phoenix Shot (RF) now hit whoever is around the TARGET, as in the original MU — before, the area was counted from the attacker, and Death Stab did not hit whoever was right behind the monster. Against groups of monsters Death Stab is stronger, and the level 1000 DK now invests in it in the master tree; in PvP that put the DK above the other classes. The matchups were re-centered by the simulation; monster damage and fight length stay on target.

Mounts 30% slower · v5.1 · 10/1/2026

• Mounts 30% slower: mounts (Dark Horse, Fenrir, Uniria, Dinorant) now move 30% slower, like the rest of movement — even with the mount hidden. The Dark Lord fights mounted on the Dark Horse and reaches range a little later, and some matchups shifted (DL against DK with +13 and own buffs went above 65%). The matchups were re-centered by the simulation; damage and fight length stay the same.

Rage Fighter and PvP with Resonance · v5 · 10/1/2026

• Rage Fighter: the flat skill bonus (vitality/10 on Chain Driver, energy/10 on Dragon Lore, agility/8 + energy/10 on Dark Side) now comes in before the target's defense, like other classes' damage. Before, part of the RF's hit did not feel any gear, neither its own nor the target's: going from +15 to +13 or changing excellent options changed other classes' damage but not the RF's, and the matchup flipped depending on the set. RF and DK matchups now stay within target with +15, +13 and 3 excellent options.
• PvP with Resonance: against players, gem Resonance is now worth less than against monsters, on both sides — it raises the attacker's damage and lowers the damage that reaches the defender. Before, it multiplied damage by up to 6 and a fight between two characters with everything at grade 10 ended in about 4 seconds. Now, with the same Resonance, the fight grows slowly with the grade: about 10 s without gems, 11 s with the 4 gems at grade 5, 12 s at grade 7 and 15 s at grade 10. Whoever has more Resonance wins comfortably, without killing in one hit: grade 10 against grade 5 wins almost every fight, in about 7 s. Against monsters nothing changes: still +500% damage and defense.
• Resonance's extra defense no longer raises SD (which only matters against players): before, each class gained SD differently, depending on its own defense. And against players the gem effects that look at life (Blood Ruby, Shadow of Kalima, Flame of Crywolf, Star of Noria, Pearl of Atlans) look at life together with SD, and Blood Ruby's life steal restores life and SD. Before, low-life classes (Elf, MG) gained much more from the Ruby than high-life ones (RF, DK): with 2 damage and 2 defense gems the Elf won 85% of fights and the RF 24%.
• What still stays outside the target, because of the gear itself: Elf against Dark Lord at level 400 with own buffs goes from 33% (+15) to 70% (3 excellent options) — the DL's hit sits close to the Elf's defense and depends heavily on excellent hits, which the 3-option set lacks. A single number cannot fit both sets; that matchup's tuning sits in the middle.

Balance with any gear · v4 · 10/1/2026

• Against players, defense now reduces each hit smoothly: a hit equal to the target's defense deals half, a hit far above it loses almost nothing, and no hit drops to zero. Before, it was hit minus defense: whoever hit close to the target's defense felt any gear difference far too much (going from +15 to +13 changed one class's damage by 0% and another's by 85%), and the excellent damage rate option decided fights. Nothing changes against monsters.
• Damage reflection (excellent items and Damage Reflect) now comes back at each class's correct scale against players. The Rage Fighter, which hits with the lowest PvP rates of the original MU, reflected about 14 times more than it should: it won fights through its armor's reflection and lost almost all of them with a set without the reflect option. To make up for it, the RF's damage against players went up.
• Result: with +15, +13 or 3 excellent options, each class's damage on monsters stays within 13% of the average (5% on the full set), and a 1v1 fight between similar characters lasts 8 to 12 seconds in all three sets. In PvP no matchup goes over 65% win rate on the full set; with +13 or 3 options, a few matchups reach 66–70% (Elf × Dark Lord, Elf × Rage Fighter and Dark Knight against DW and Summoner): in those pairs the gear itself changes who wins, by 28 to 39 points, and their tuning sits in the middle.
• In the simulation, PvP now also uses mana potions (as in hunting), and the rotation search is more thorough: each class's 5 strongest skills are tried first and the DK tries every combo. The DK that looked about 25% weaker with the 3-option set was the simulation picking a worse combo (with Blood Storm, which has a long cooldown), not the character.
• The automatic balance every 3 days now also tunes party buffs (damage on monsters and each matchup in a party) and only applies an adjustment if no target gets worse in any of the three gear sets.

Exact attack speed · v3 · 9/29/2026

• Every character now hits at the exact pace of their own attack speed (or magic speed, for magic skills). Before, the server rounded each hit to 250 ms blocks: an 800 ms interval became 1000, and 5% more speed sometimes changed nothing and sometimes gave 25% more hits. Now every point of agility, excellent speed option or speed buff counts in the right proportion.
• Because of that each class's damage changed (whoever fell on the worst rounding gained hits), and the balance was redone from scratch by the simulation: damage per second on monsters equal across the 7 classes (within 5% of the average) and 1v1 fights of about 10 seconds (8.8 to 11 s), at levels 400 and 1000, without buffs, with their own buff and in a party. The game's average damage stays the same.
• With the 3-option excellent set, at level 1000 the DK is about 25% below average and the DW about 15% above: the DK's Master Level damage depends heavily on excellent damage, and that set lacks the excellent damage rate option. At level 400 the 3-option set is now within the monster damage target.

Party and gear variations · v2 · 9/29/2026

• In a party, with the other classes' buffs (Elf, DL, DK, RF…), all 7 classes keep the same damage per second on monsters and PvP stays balanced — before, a buffed Rage Fighter won almost every fight.
• Multi-hit skills (Rage Fighter): the flat bonus from buffs (Greater Damage, Critical Damage…) now counts once per use, split across the hits, instead of once per hit.
• Two weapons (DK with two swords, Summoner with staff and book): the weapons' excellent speed, damage and excellent-damage-rate options count the average of both hands, not the sum. Before, the DK depended on those options much more than other classes. The DL's Dark Raven follows the same damage adjustment as the DL.
• The balance now also holds with other gear (+13, and excellent with other options), not only the test set. Against players, damage reflection from excellent items is worth 20% (5 pieces with reflection = 5%); against monsters it stays in full.

First class balance · v1 · 9/29/2026

• All 7 classes now deal the same damage per second to monsters (within 10% of the average) at level 400 and 1000, with and without their own class buff. The game average stayed the same: those above came down, those below went up.
• Rage Fighter: skills hit several times per use again, as in the original MU (Chain Driver, Dragon Lore and Phoenix Shot 4 hits; Large Ring Blower, Upper Beast and Dark Side 2). Before only one hit came out and the RF dealt a quarter of what it should.
• PvP: the SD shield no longer recovers mid-fight — only 5 s after the last hit exchanged. Before, it came back faster than it went down and the fight became a slow race of life.
• PvP: a 1v1 fight between similar characters lasts about 10 seconds, and each class matchup was tuned so none has a big edge over another. Damage reflection now goes through the attacker's SD, like a hit.
• Class buffs: still full against monsters; in PvP, Mana Shield, Damage Reflect and the damage of whoever has their own buff up were tuned so the fight stays balanced with and without buffs. The Summoner's Berserker moved part of its bonus into the class's base damage.

Simulation: tools/dev/sim-balanco.ts (real server logic); tuning: tools/dev/ajusta-balanco.ts; goals locked in server/src/game/test-balanco.ts.

Code reference: shared/src/balanco.ts · shared/balanco-dados.json · server/src/game/balanco-ciclo.ts · server/src/game/combat.ts → rollPlayerDamage, rollPvPDamage · tools/dev/sim-balanco.ts