The roster question people are really asking is which few to learn first, not how many exist.

Short answer

Take the top of the list as a shortlist, then pick whichever matches how you actually play.

Pick rate and win rate disagree constantly, and the gap between them is the interesting part. Seasonal balance passes tend to overcorrect first and settle later. The useful part of any answer is the part that tells you what to do differently.

What the last patch moved

A difficult pick with a high ceiling is worse than an easy one until you have put the hours in. The in-game roster screen is authoritative; community lists lag it by days at minimum. Nothing here assumes a particular playstyle beyond wanting the thing to work. Role coverage inside a team matters more than the individual rating of any one pick.

Roster additions shift everyone else even when nothing about them was touched. It helps to understand why a step exists, because then you can adapt it when your situation is slightly different. The strongest character and the right character for you diverge more often than lists admit.

Picking two rather than eight

Anyone described as beginner-friendly is usually easy to play and hard to play well. A specific answer beats a comprehensive one when you are in the middle of something. Balance passes usually move numbers rather than reworking a kit, so the character still plays the same way.

  • Watch one match of the pick before committing time to it.
  • Read the rework notes rather than a summary of them.
  • Ask what a step is meant to achieve before deciding to skip it.
  • Compare pick rate against win rate rather than trusting either alone.

Where to start if you are new

Tier lists describe a skill bracket, and most of them do not say which one. Where something depends on your platform or region, that is called out rather than glossed over. The strongest pick in professional play is often the worst pick in a solo queue. Difficulty of execution is missing from most rankings, which is why some entries look underrated.

Where two approaches both work, the one with fewer moving parts is the one to keep. A list built from professional play describes a game most players are not playing. New additions are typically overtuned at release and corrected within two patches.

Good places to start

Half of what gets described as a bug is a design decision people dislike. A character who dominates at the top level can be genuinely bad everywhere else. Counters matter more than raw strength in anything with a draft or a mid-match swap.

Nothing below assumes you have read anything else first. Cosmetic reworks and balance reworks get announced together, which is why patch notes read confusingly. A rework can move someone across the entire list without changing anything about how they look.

How the roster breaks down

The interesting question is usually why the obvious approach fails. Every roster has one entry that is quietly excellent and permanently unpopular. Learning two well beats learning eight badly, and the second should cover the first one's weakness.

Where people go wrong

  • Downloading a helper tool for something the game already does in its own menu.
  • Skipping the prerequisite because it looked optional in someone else's summary.
  • Ignoring the platform-specific note and then wondering why the steps do not line up.
  • Rushing the last step, which is usually the one with the actual requirement in it.

BG3 FAQ

Do I need any mods or third-party tools for this?

No. Everything described here works with the base game on any supported platform.

Does this differ by region?

Occasionally. Where it does, it is usually storefront or server behaviour rather than the game itself.

Is there a faster way to do this?

Usually, and it usually trades reliability for speed. Which one you want depends on whether you plan to repeat it.

How long does this take?

Minutes if it goes cleanly, an evening if you hit the edge case in the caveats list. Neither is unusual.

If you only take one thing from this: the answer near the top is the part that matters, the rest is context.