The current verdict: unranked
There is no defensible S-to-D ranking in the original-game evidence we reviewed. The launch announcement describes alpha balance, and the public class comments do not supply consistent build identifiers, progression levels or run conditions. Assigning tiers would imply a precision those sources do not support.
Evidence reviewed October 15, 2026. The original selector was inspected at v0.91.0 (0d319264), but no controlled playtesting or win-rate dataset underlies this page. “Unranked” means insufficient comparable evidence, not poor performance.
For a practical starting point, choose the class you want to learn and keep the early conditions consistent. The beginner guide explains what to record before deciding that one class is stronger.
What the original-game reports say
| Class | Public report | Evidence limit |
|---|---|---|
| Mage →Unranked | A player reports strong performance; another describes needing to kite. | First-person anecdotes; version, gear and build unspecified. |
| Warrior →Unranked | A commenter relays a friend’s positive assessment. | Second-hand report; no recorded run or test conditions. |
| Warlock →Unranked | A player says movement was necessary to avoid dying. | Positioning feedback, not a comparative damage or survival test. |
| Paladin, Hunter, Rogue, Priest, Shaman, DruidUnranked | Their selection-screen information was observed directly. | Starting values establish a reference, not relative combat strength. |
These reports come from the original launch discussion ↗. They are useful questions to investigate, not a consensus ranking. See the linked class notes for how to evaluate each claim.
Verified starting values, not strength tiers
The original character selector provides a useful baseline for all nine classes. A higher displayed health value does not by itself establish a stronger class: starting spells, race, upgrades, recovery and player decisions also matter.
Character selection inspected October 15, 2026, with empty equipment, 0/51 talent points and Normal difficulty. Health values are what the selector displayed, with the selected race recorded alongside them. They are not measured combat health, DPS or a ranking.
| Class | Starts with | Displayed health | Selected race |
|---|---|---|---|
| Warrior → | Cleave | 230 | Human |
| Paladin | Consecration | 245 | Human |
| Hunter | Auto Shot | 240 | Dwarf |
| Rogue | Sinister Strike | 190 | Dwarf |
| Priest | Smite | 170 | Dwarf |
| Shaman | Lightning Bolt | 195 | Orc |
| Mage → | Fireball | 215 | Undead |
| Warlock → | Shadow Bolt | 165 | Undead |
| Druid | Wrath | 205 | Tauren |
Character-selection UI, v0.91.0 (0d319264), inspected October 15, 2026. All nine class tiles, displayed health, starting spells and selected races were read directly. No full combat test was completed.
How to compare classes fairly
A meaningful tier list needs a version, a goal and repeatable conditions. We would assess opening survival, wave handling, boss completion and consistency separately before combining them. A comfortable first run and a strong late-progression setup answer different questions.
- Identify the game. Record original browser or local build, version if available, and test date. Do not merge community-reconstruction results into the original dataset.
- Match the starting conditions. Keep map, mode, permanent equipment and Talent XP investment comparable. Note any differences you cannot control.
- Record the offers. Upgrade randomness can change the outcome. Save the choices, not just the final survival time.
- Repeat the comparison. Keep successes and failures. One standout run does not establish consistency.
- Explain the recommendation. Publish the number of attempts, relevant results, limitations and the player goal the ranking serves.
The selector provides verified starting information; the early community comments add limited context about Mage, Warrior and Warlock. Neither establishes consistent comparative outcomes, so all nine classes remain unranked here.
Keep a record of your next run
Use this small notebook to make your next comparison easier. It saves one current note locally; copy it somewhere durable before replacing it. The nine class names match the original selector observed in v0.91.0.
Original game vs. community reconstruction
The OutThisLife repository ↗ describes itself as a from-source local recreation and credits Aethus. Its code and README can document that project, but do not establish the original game’s current balance. We have not used its spell values, health pools or controls to rank the original.
Evidence & update policy
This page is dated by the evidence review, not an invented patch number. Rankings should only change when comparable original-game results or identifiable version changes justify them. No scheduled update or live patch feed is implied.
Primary source for browser play, progression and the community reports cited here. Community comments do not identify their build numbers. Original selector observed separately at v0.91.0 (0d319264).
Community reconstruction README ↗ — used only to identify the separate project, not to rank the original.