Class Tier List
A class can be top tier in raids and bottom tier in abyss. One letter cannot say that, so these are rated separately.
The Korean ratings on this page were read during the week Korea was buggy. From the 8/13 "double rune options" event, some damage runes and some class passives applied twice while other skills ignored them entirely. Korean players only confirmed it was a bug on 8/18–8/19, and the 8/20 update closed it. Measurements taken in between are partly measuring the bug rather than the class. The Korean ordering below was read inside that window — treat it as pending a re-read against the post-8/20 game. Taiwan has not entered Season 2, so this does not affect the Taiwanese service. 2026-08-13 → 2026-08-20 · ruliweb 300143_76360300, _76362650, _76369387, _76369559 (2026-08-18 → 08-19); closed by the 2026-08-20 update, whose notes list the affected runes
TW runs the Mars abyss and these ratings are from that season. KR shows Runda (S2) — the two ladders disagree.
Only content TW has now — TW runs ~460-480 days behind KR, currently around KR early-May-2025
Self-heal + stability are at the highest level, and there is no need to select a position, so it can be completed even in idle auto mode — “If you just leave it to it, it will take care of itself.”
Launch era · Source: inven/6366/203; vortex/528051
Auto-optimized due to simple operation with wide area floor + infinite taunt tank. However, skill auto 4 must be turned off in the hunting field (infinite loop as Jangpan keeps respawning)
Season 1 · Source: inven/6366/203; dc/enban/267678; namu 빙결술사
With long-distance rapid fire, the speed of clearing out random mobs is fast and there is a lot of evasion so Auto doesn't die.
Season 1 · Source: inven/6366/203; dc/enban/267678
“The best self-healing device in the game” + No. 1 in long-term maintenance with super armor. Counterexample: S0 complains that in low-level fields, the movement is slow and the mob dies before the skill is used.
Launch era · Source: dc/enban/267678; dc/mabinogimobile
Wide-area damage that only requires a cooldown from a distance — only skill number 5 must be removed from the auto list to be effective
Season 1 · Source: dc/enban/267678
Maintaining the stack is virtually a continuous button press, so there is no room for the auto AI to ruin the cycle — 1st place in the thief class auto (Double>Rogue>Fighting)
Launch era · Source: inven/6366/216
Excellent auto performance due to self-maintenance (2nd recommended for beginners). Flaw: Auto AI does not use 3 (Life Pulse) often.
Season 1 · Source: dc/enban/267678; ruliweb/72224884
No auto-specific evaluation — the only comment was that the S1 “loses a lot of power.” Currently, the S tier on our site is derived from raid standards, not everyday standards.
Season 1 · Source: dc/enban/267678
Among the Bard series, Otto is the only one that rolls — “The best character in the Bard series” is not absolute performance, but relative evaluation within the series.
Season 1 · Source: dc/enban/267678
It goes back to auto, but "you can't expect a big deal" — CONTESTED, Namuwiki demands manual operation, saying you can't move during casting and die when auto.
Season 1 · Source: dc/enban/267678 vs namu 화염술사
It's a preheating type that stores and explodes poison stacks and overcharges, but the auto AI just lets the cooldown go + the cycle itself doesn't roll without a specific weapon rune.
Season 1 · Source: inven/6366/216; dc/enban/267678
The general content is painful due to the lack of a wide area (“not recommended because it feels dirty”), and because it focuses on linking, it has poor compatibility with Auto — the exact opposite of the top level of end content.
Season 1 · Source: inven/6366/216; dc/enban/267678
Auto was hit as his movement was blocked while casting a skill ("Auto's asshole") — The reason for the existence of this file is that he was Tier 0 in the raid at the same time.
Season 1 · Source: dc/enban/267678
Single specialization + focus on slow strikes, low throughput of minor mobs and lowest auto rating
Season 1 · Source: dc/enban/267678
Tanks and dildos are mediocre, and “autos are also terrible, so freezing is good for your mental health.”
Season 1 · Source: dc/enban/267678; vortex/528051
Auto AI inserts basic attacks between skills, completely eliminating the effect of the raid rune — the most concrete evidence in the corpus that “AI can’t use kits.”
Season 1 · Source: dc/enban/267678
Advance Deal, Casting, Charging, Preheating 4-Use + Auto is meaningless as you have to manually pass the evil image threshold
Season 1 · Source: dc/enban/267678; dc/mabinogimobile/622227
It's a warm-up type, but if you move frequently, your DPS drops significantly — "It's very powerful in theory, but you can never put that theory into practice."
Season 1 · Source: dc/enban/267678
Why the rankings differ per content
These mechanics decide the letters, and are worth more than the letters.
📄 Cells still shown in Korean or Traditional Chinese quote the primary-source records (Korean community / Taiwan official) verbatim — they are labeled, not passed off as translations.
Damage formula term G: a broken (defenseless) target takes (1+break/5250)×(1+defenseless damage%) PLUS a flat 20% base plus break-type tag bonuses (strike/mental break = hard hit+10%/multiple hit+10%; stun = multiple hit+20%). All major formula terms A~L MULTIPLY, so damage crammed into the window compounds against everything else you have running. · BURST end: Fighter·Electricist·Great Sword Warrior(Berserker Spear)·Priest(Judgment Banking)·Archer Archer. SUSTAINED end: Dark wizard, crossbowman, longbowman, monk. · Decides RAID and ABYSS. Irrelevant to trash clearing (mobs die before a break matters).
dc/enban/134; dc/mabinogimobile/1160569 (2025-06-04); dc/mabinogimobile/887412 (2025-05-17)
Extends the break window AND raises damage taken. Held by exactly THREE classes in the whole game: 전사, 기사, 빙결술사. Diminishing returns if several people fire it on the same target and total break time is capped ⇒ one 브익 seat is guaranteed, a second is wasted. ⚠ S2 NUMBERS ARE IN FLUX —: the 2026-07 formula post has 브익 multiplying the break term and tags by ~1.5×, while a 2026-06-25 patch-reading post says the S2 values were HALVED (무방비 20→10%, 강타/연타 10→5%) with duration raised instead. · HOLDERS (warrior/knight/frostmage) are a guaranteed party slot: "Having a break extend is enough to secure a spot in the party." But the same tank design gives them long cast times and a LOW personal ceiling: "Freezemancer's own burst firepower and damage ceiling are low". Article is the fastest applier — a measured break rush setting cut time-to-break 29s→16s at the cost of DPS 10.6M→6.5M. · Decides RAID and ABYSS party composition, and decides them DIFFERENTLY. Worth ~zero solo and ~zero in daily trash.
namu 기사·빙결술사 (two pages agree); dc/enban/134; dc/mabinogimobile/2169437, /2170776, /2170833; mabilife 어비스 티어
Only the hit that FILLS the gauge counts as brake; everything after is defenseless. Break types (Blow / Mental / Stun / Charm / Freeze / Fall) each grant different tag bonuses and do not stack — "Only the damage increase attribute of the break type applied at the moment applies", and re-breaking WIPES the previous tag. “If the blow brake that most people have is applied, the other brakes won’t work.” So in an 8-man, a random DPS filling the gauge first can LOCK OUT the party's Freeze Break or Stun Break. · Providers of the good types: Healer (Mental), Crossbowman (Stun), Musician (Charm), Freeze Wizard (Freeze), Warrior/Swordsmith (Blow). Loser: Musician, whose break amount is RANDOM 1-3 spaces and therefore cannot be scripted around. · Pure RAID/ABYSS party-composition axis. It is the mechanical reason KR players argue about WHICH class lands the break, not just how much damage each does. · dc/mabinogimobile/1160569 (2025-06-04)
S2 season that trigger on Break (Freeze Wizard Icicle Shackle, Penetrating Sword Chamber) do NOT fire in a defense-only state, and do not fire at all on break-immune bosses — "It's not a break, it only deals defenseless damage!", "You have to play as season 0 because the season skill is sealed." Break-immune bosses are confirmed to exist (Runda Caverns of Madness boss Deathpel). · Hurts exactly the classes whose S2 kit is built on break: Freezer, Warrior. A break-immune boss disarms the break trio outright. · S2 ABYSS/RAID only. TW does not have S2 season, so this trap does not exist on TW.
arca/177368137 (2026-07-19, via agent browser); namu 빙결술사
Counting break-capable: Warrior 4, Freezer 4, Longbowman 2.5, Monk 2.5, Fighter 2.5, Archer 2, Pyromancer 2, Bard 2, Musician 1-3, and 1 each for Greatsword Warrior, Wizard, Shockman, Crossbowman, Dancer, Rogue, Dual Blade, Priest. A separate community ladder ranks break-APPLYING skill: Tier 1 Swordsman (Warrior)·Bard / Tier 2 Longbowman / Tier 2.5 Swordsman / Tier 3 Crossbowman·Mage·Freeze Wizard·Musician / Tier 4 Dancer·Monk·Healer·Archer·Great Sword Warrior / Tier 5 Pyromancer / Tier 6 Priest (friendly link). · BEST: Warrior, Bard, Freezer, Longbowman. WORST: Priest, Pyromancer, Greatsword Warrior, Archer. · Decides SOLO/4-man ABYSS hardest — in a small party nobody else covers your break debt. In 8-man raid someone else always breaks, so it matters less. · dc/mabinogimobile/1496175 (2025-06-25); dc/mabinogimobile/476721 (2025-04-22)
Some kits need a stack or resource runway before they do their real damage. Rogue: poison stacks + overcharge, first explosion is weak, "poison explosion damage returns to normal from the second damage cycle." Shockmaster: Overcharge stacking, longest warm-up in the game (mabilife: ~40s) and probability-gated ("Casino Deal"). Dark Magician: Encroach → Possession state machine. Musician: first damage + casting + charging + preheating all at once. · NO WARM-UP: Greatsword Warrior (mabilife singles this out as a real abyss advantage), Archer, Freezer. LONG WARM-UP: Lightning magician, dark magician, thief, musician. · Punished twice — in DAILY (trash dies before you ramp) and inside short break windows. Rewarded in long single-boss fights and in raids whose boss breaks rarely.
namu 도적·전격술사 (snippets); mabilife 어비스 티어; dc/enban/267678
namu stats states outright that multi-hit stats are not worth investing in because "fight against one boss in final content such as Abyss, Raid, etc." Endgame is one target; daily is a screen of trash. Crossbowman is "a single-target sustained dealing specialist rather than a pure AoE dealer"; Longbowman's AoE skill exists but its single-target loses to Deadly Shot with a longer delay. · SINGLE-TARGET SPECIALISTS: Longbowman, Crossbowman, Blitzkrieg, Fighter (Martial Artist explicitly "absence of AoE"). AoE/CLEAR SPECIALISTS: Freezer, Pyromancer, Mage, Greatsword Warrior. · Inverts between DAILY and RAID/ABYSS. This alone explains why the fighter sits near the top of every endgame list and gets not recommended for daily grinding.
namu 능력치·석궁사수·장궁병 (snippets); dc/enban/267678
Abyss Hell difficulty applies potion restriction as a fixed modifier, so self-heal stops being a convenience and becomes throughput. Greatsword Warrior has "the best self-healing skill in the game" plus super armor → "capable of stake damage"; mabilife credits it with "economics" (lower potion burn) in abyss. Dancer's S2 step rune adds self-heal. Monk's heal is "at the top of the game". · SELF-SUFFICIENT: Greatsword Warrior, Monk, Healer, Freeze Wizard, Rogue (Steak Aggro Reset). FRAGILE: Dualblade ("very few survival skills"), pyromancer, crossbowman/longbowman (rooted casts). · Decides ABYSS Hell and unattended DAILY auto-play. Nearly irrelevant in 8-man raid where 2 healers are the meta.
; dc/mabinogimobile/529792; mabilife 어비스 티어; inven/6366/216
Some kits multiply by headcount. Dancer's Applause: 4 extra hits on self, plus 2 extra hits on up to 7 allies. Bard's ult Festival Song: party-wide invulnerability + party ATK +30% + cooldown refund per tick. Both are worth nothing solo and enormous at 8. · SCALES WITH HEADCOUNT: Dancer, Bard, Priest, Healer, Musician (debuff). FLAT REGARDLESS: Fighter, Dual Blade, Crossbowman, Longbowman. · 8-man RAID vs 4-man/solo ABYSS. namu on Dancer: "In 8-player raids, it's comparable to a top dealer" — the same class every flat list puts dead last.
namu 댄서 ; namu/inven 음유시인 궁극기;
Abyss gates progression on a 10-minute time attack per stage (shorter on hell), and hell adds a clear-time ranking — the community calls record-chasing Gigaeng (record-breaking). The problem with namu says plainly that “jobs with low damage are discriminated against or excluded from Time Attack content.” COUNTERWEIGHT, and we must publish it: "If you build up the specs for any class, there will be no problem unlocking up to Abyss Hell 2 difficulty" — every class can CLEAR; only the leaderboard is exclusionary, and time-attack rewards are one-time. · EXCLUDED FROM GIGENG PARTIES: low-burst classes. REQUIRED: a break extend holder + top burst. · ABYSS only. Raids have a weekly clear, not a leaderboard; daily has no clock.
namu 문제점·어비스 (snippets); line 86 (TW 10-min limit, official); arca/164217344 (403, via snippet)
The auto AI plays a fixed priority list. It cannot bank cooldowns, cannot cross a resource threshold on purpose, and interleaves basic attacks. Documented failures: Bard — the AI inserts basic strikes between, which erases the raid rune effect entirely; Thief — “Because I can’t use my skills to operate”; Martial Arts — “It’s mainly about collaboration, so it doesn’t have much compatibility with Otto”; Crossbowman/Longbowman — rooted casts eat hits; Healer — AI under-casts skill 3; Wizard — only good if you EXCLUDE skill 5 from the auto list; Freezer — auto-casts skill 4 whose radius+duration re-pulls respawns endlessly, so namu says turn skill-4 auto OFF in Hunting Grounds. Auto is only viable for Quest / Day/Silver Coin Dungeon / Hunting Ground — "From deep dungeons onwards, it is not easy to run in full soloing auto." · AI-PROOF: Monk, Freeze Wizard, Dual Blade, Greatsword Warrior. AI-BROKEN: Bard, musician, thief, fighter, lightning magician, dark magician, crossbowman, longbowman. · DAILY only. Zero bearing on raid or abyss, which are played manually.
dc/enban/267678 (2025-10-27, 20-class evaluation); inven/6366/216; namu 사냥터·빙결술사 (snippets)
Formula term L: if your magic resistance is below the zone's magic pressure your damage is multiplied by 0.5^(deficit/1000) — exponential collapse; above it you gain 1+0.4×(1−0.5^(surplus/10000)), i.e. up to +40% final damage. Runda Abyss Very Hard needs Combat Power 76,000 + Resistance 2,200; Kabrak needs ~65,000 + ~2,000. · Nobody — it is a GEAR check, not a class trait. Its only class effect is indirect: it must be satisfied before any class difference is visible at all. · Gates S2 RAID and S2 ABYSS entry on KR. TW does not have this system at all.
dc/enban/134;
The same class swings by boss. Tabartas raid applied break only once per fight, which the community flagged as "Sustained dealers" (favors sustained dealers). Abyss is the opposite: "Bosses are also prone to breaks" — which is precisely why Freezer is rated Tier 0 there. mabilife notes Pyromancer "Contaminated Waste Site" struggles because that fight's break windows are short. · Same class, opposite verdict, depending on the fight. This is the strongest single argument that ONE tier per class is structurally wrong. · Explains RAID-vs-ABYSS divergence and even RAID-vs-RAID divergence.
; namu 빙결술사 (snippet); mabilife 어비스 티어
Where sources disagree
Not averaged away — who said what, so you can judge.
Priest vs Healer — Who is Favored on Hard Difficulty?
What this site said: Uri and mabilife series: Priest is preferred at high difficulty due to burst heal + damage
What KR sources say: KR's original text was opposed for a long time: "Healer = dot damage, dot healing, multiple notarization" vs "Priest = one-shot damage, one-shot healing, single notary" → Healer with multiple notaries is the raid pick, and the priest is the solo pick. In S1, Priest "The party preference is not high. Not recommended" / Healer "The party preference is high, so it is 2nd recommended for beginners." In S2 2026-08, a user who has been the main priest since Season 0 said, “I am the only priest in the guild,” and “It is a burdensome position to be in.”
Verdict: The sim/tier numbers say that S2 priests are the strongest supports, and the actual party invite experience still says healers. Print the two sentences side by side. Our "Priest Preferred" Sentence Should Be Relaxed Until Reconfirmed
Source: HIGH (양측 모두 원문 확보)
Freezer - “Abyss Top” or the lowest dealer?
What this site said: mabilife #1 Ranker Abyss Tier 0 “You can’t even start without it”; 1st place overall in S2 Hell 1 ranking score; mabinogi.dev "In fact, the best"
What KR sources say: 2025-04-07 Ranked 12th out of 15 in Average DPS table; namu Disadvantages: “Because single damage is not excellent, performance is relatively disappointing in boss battles”, “DPS is low”, overall review: “A wide area specialized tank”
Verdict: It's a category error. Both are correct. Ranked 1st in party value and BUIK utility, but ranked 1st in individual DPS. dc/enban/913135 does not count Freezing as a dealer, but classifies it as the number one tank. mabinogi.dev cannot be obtained directly from the text (797 to 944 bytes of response) and appears to be an SEO/AI aggregate, so do not cite it as a community consensus.
Source: HIGH
Freezer - Abyss vs Raid (same class, opposite conclusion)
What this site said: Abyss: "Bosses are also prone to breaks, so the effectiveness of Break Extend is also amplified, which has a significant impact in drastically reducing dungeon clearing time."
What KR sources say: Raid: “On the other hand, raid bosses do not have much influence because they only break at certain times.”
Verdict: This is not a contradiction, but the reason this file exists. The same document separates two pieces of content in one sentence — the neatest quote about why one class is impossible.
Source: HIGH
Greatsword Warrior — Is he the strongest in maintenance or a nuisance to the party?
What this site said: namu: The only one that can deal with stakes in the writing push gimmick, and can even play solo in an 8-player raid; mabilife Abyss Tier 1 “So strong that there is almost no need to use potions.”
What KR sources say: namu itself writes that it is "a top priority for parties focused on breaking records on Hell 15 difficulty"; Developer 2026-08-12 "Damage dealing ability tends to decrease in party play."
Verdict: Divided by content: Best in clear/survival content, avoided in record-breaking parties. TW corresponds to the former, so it is strongly recommended based on TW.
Source: HIGH
Developer vs Community (S2, 2026-08-12) — Content not on TW site
What this site said: Developer: Lightning Master "Confirmed to be quite high-performance", Warrior "High-ranking class", Swordsman "Rank or DPS is not low"
What KR sources say: Immediate backlash from the community: “What the hell is sword fighting like? Haha, I understand why these bastards have such shitty balances”, “I was so stupid that I brought a crossbow and it hit me hard”, “In the end, the day will come when sword fighting will be at the bottom.”
Verdict: Load both. The fact that Nexon’s judgment and user experience are at odds is itself information. No TW site carries this
Source: HIGH
Bard ultimate name
What this site said: Our repo and some secondary sources: 「Song of the Festival」
What KR sources say: namu original text and KR gallery: The name of the ultimate skill is Battle Symphony. In the gallery title + text search, “Festival Song” was found 0
Verdict: Correction to “Battle Symphony”. 「Festival Song」 appears to be a TW/CN localized name or a typo — do not use it on the KR label page. The effect (invincibility + ignores ally breaks + attack power 30% + party cooldown reduction) is confirmed in both cases.
Source: HIGH
S2 figure for break extend
What this site said: dc/enban/134 Damage formula (updated 2026-07-19): Break term and tag damage become approximately 1.5 times stronger in break state.
What KR sources say: dc/mabinogimobile/2169437 (interpretation of 2026-06-25 patch): Break value in S2 halved — defenseless 20% → 10%, hard/multiple hits 10% → 5%, continuous 5% → 2.5%. Instead, duration/uptime increases.
Verdict: Don't post numbers. Post only the role — the description of S2's Beek's change in character from a "d deal amplifier" to a "window length provider" is compatible with both sources.
Source: MED
Warrior — bottom of 10,000 years or top of S2?
What this site said: 2026-04-05 Ranked 18th/C in the tier table, lowest dealer. Our site is also C
What KR sources say: Developer 2026-08-12 "The numbers are as good as balanced. It is a top class."; 2026-08-14 2nd place in 4-person Abyss ranking tank, marked as tier 0 tank in some communities
Verdict: It is highly likely that it actually flipped between S1 and S2. However, this cannot be confirmed as there is no teardown dedicated to S2. Since there is no S2 in TW, S0/S1 evaluation (lowest net deal + profit utility) must still be used based on TW.
Source: MED
Longbowman — 2nd to the Scarecrow, but rejected by the party
What this site said: 2026-08-14 Scarecrow DPS 2nd (58.62 million); Developer: “The numbers are as expected.”
What KR sources say: At the same time, “I’m a bit weak, so even hell doesn’t help me”; arca "Some people say longbows are shit, others say they're strong, which is correct."
Verdict: Scarecrow DPS ≠ Textbook case of party value. This will serve as proof of why we need to show “raid DPS bench” and “party preference” separately.
Source: HIGH
Dual Blade
What this site said: Some S2 Tiergle Tier 1; 2025-10 S Tier
What KR sources say: 2026-08 “Double Abyss is overwhelmingly last place haha… Not only the performance but also the perception is shit”; namu "Dual blades, which are only good at ground damage, have relatively low ratings."
Verdict: unresolved. Since they are overturned within 48 hours, neither side should post anything and only mark it as contested.
Source: MED
Pyromancer Auto Performance
What this site said: dc/enban/267678: Returns to auto (no deal expected); vortex is S tier
What KR sources say: namu: Since you cannot move while casting, you will die from being hit by a mob on auto mode, so manual operation is required.
Verdict: Unresolved three-way match (third party drhot552 excluded due to AI slop risk). Attaching contested badges from Otto Marks
Source: MED
Fighter S1 Tier
What this site said: 2026-04-05 5th place/A “Famous Dungeon Dealer”
What KR sources say: 2026-06-08 Tier 4/Bottom
Verdict: 2026-02 Caught between the heavy-hitting nerf and the Beek meta-checking design — this can be explained by the difference in perspective. Show both dates
Source: MED