How we score

The idea

A council of independent AI models reads the same fact pack: patch notes, class research and player reports. Each model scores every class from 1 to 10 for Leveling, Endgame PvE and PvP. We average the scores. No single model decides.

The formula

Score = average × 10, so an average of 8.1 is 81%. Overall = (Leveling + Endgame + PvP) ÷ 3. The overall score sets the tier.

TierScoreQuality
S80 and upLegendary
A70 to 79Epic
B60 to 69Rare
C50 to 59Uncommon
DUnder 50Poor

Agreement marks

Next to each class you may see how much the models agree. We look at the biggest gap between any two models on one axis.

  • Models agree Biggest gap is 2 points or less.
  • Mostly agree Biggest gap is 3 or 4 points.
  • Models split Biggest gap is 5 points or more.

What is not in the score

The Forever Open tournament is not counted. It was level 30 only and ran in launch week, so it says little about leveling or endgame. Player votes from the Elders addon will be added later.

The council

The council scored the classes:

  • beam
  • big-pickle
  • deepseek-v4.1-flash
  • glm-5.3-flash
  • longcat-2.5-preview
  • mimo-v2.6-flash
  • nemotron-3-ultra
  • inception_mercury-2.5
  • nvidia_nemotron-3-ultra-550b-a55b
  • poolside_laguna-s-2.1
  • thinkingmachines_inkling
  • xingchen-agi_xing4.0-29b-a4b
  • space-bunny

The council scored the 27 specs, from the same kind of fact pack:

  • beam
  • big-pickle
  • deepseek-v4.1-flash
  • glm-5.3-flash
  • gpt-6-luna
  • longcat-2.5-preview
  • mimo-v2.6-flash
  • nemotron-3-ultra
  • nemotron-3.5-lightning
  • nemotron-3.5-super-vl
  • inception_mercury-2.5
  • nvidia_nemotron-3-ultra-550b-a55b
  • poolside_laguna-s-2.1
  • thinkingmachines_inkling
  • space-bunny

Left out: liquid_lfm-2.5-2.6b. Reason: too small; mislabeled classes.

Every model's scores

Scores 1 to 10 from each model. Avg is the mean of the valid scores.
Classbeambig-pickledeepseek-v4.1-flashglm-5.3-flashlongcat-2.5-previewmimo-v2.6-flashnemotron-3-ultrainception_mercury-2.5nvidia_nemotron-3-ultra-550b-a55bpoolside_laguna-s-2.1thinkingmachines_inklingxingchen-agi_xing4.0-29b-a4bspace-bunnyAvg
Paladin998879989786—8.1
Mage68897776786767.1
Hunter109910910991099799.2
Warlock88798888868777.7
Shaman76886776777676.8
Warrior55445465685555.2
Priest44545644553244.2
Rogue56575556565655.5
Druid44434444553343.9
Spec scores 1 to 10 from each model, best spec first.
Specbeambig-pickledeepseek-v4.1-flashglm-5.3-flashgpt-6-lunalongcat-2.5-previewmimo-v2.6-flashnemotron-3-ultranemotron-3.5-lightningnemotron-3.5-super-vlinception_mercury-2.5nvidia_nemotron-3-ultra-550b-a55bpoolside_laguna-s-2.1thinkingmachines_inklingspace-bunnyAvg
Retribution109910101010989910910109.5
Protection99881088989999998.7
Frost98989989899109998.8
Demonology99991098989999998.9
Discipline6676776747678676.5
Enhancement9998999988998998.7
Holy8887877878888887.7
Protection8776877777787777.1
Beast Mastery1010101010101010910910910109.8
Feral Combat6776866757676686.5
Arcane6676767657667676.3
Destruction7677776757677686.7
Restoration4454444535755354.5
Survival8777878868788787.5
Marksmanship7767867757677766.7
Subtlety6665776757666676.2
Restoration5554455535555454.7
Fire7677877757677776.8
Arms6656566566556565.6
Balance5555655646565565.3
Affliction9999988988898998.6
Elemental5555555555555555.0
Holy5565665656566565.5
Assassination6566666656466655.7
Combat6676676758577676.4
Shadow4454545544445544.4
Fury4444445436334444.0

Sources

Forums

Trackers

Patch notes

Tournament

Downloads