Arknights: what Elite 1 and Elite 2 actually buy you
Promotion is the most expensive decision in the game and its raw-stat return shrinks at every step. Measured across all 126 six-star operators, here is what each phase is really worth.
Raising an Arknights operator has three phases, and they do not cost anything like the same. Elite 0 levelling is cheap. Elite 1 is a real investment. Elite 2 is the single most expensive thing most accounts do.
The raw-stat return runs in exactly the opposite direction.
The measurement
Every 6★ operator's growth curve is a set of breakpoints on a continuous axis: Lv.1–50 is Elite 0, Lv.51–130 is Elite 1, Lv.131–220 is Elite 2. Below is the median percentage gain within each phase, across all 126 six-stars whose curves span the full axis.
| Stat | E0: Lv.1→50 | E1: Lv.51→130 | E2: Lv.131→220 | Lv.1→max |
|---|---|---|---|---|
| ATK | +43% | +32% | +21% | +130% |
| HP | +35% | +30% | +28% | +131% |
| DEF | +46% | +32% | +21% | +132% |
A fully-raised six-star has roughly 2.3× its Lv.1 numbers. More than a third of that arrives during the cheapest phase of the three.
Note also that HP behaves differently from ATK and DEF. Its gain is nearly flat across the phases (+35 / +30 / +28), while ATK and DEF fall off a cliff at E2 (+43 / +32 / +21). A promotion to E2 buys a defender proportionally more survivability than it buys a guard damage.
So why is E2 worth it?
Because you are not buying the numbers. The stat curve is the visible part of a promotion and the least important part of it. E2 is bought for the third skill, the second talent, the module slot and the higher skill mastery ceiling — none of which appear on a stat line, all of which change what the operator can do rather than how much of it.
This is a specific case of a general problem with reading stat tables, set out in how to read a stat percentile: a statline cannot see a skill, and in Arknights the skill is usually the operator.
What follows for build order
Breadth beats depth early
If you are choosing between taking one operator to E2 and taking three to E1, the raw-stat arithmetic favours breadth by a wide margin — you are collecting the +43% and +32% bands three times over instead of paying the most for the +21% band once. The counter-argument is real and skill-shaped: some operators only become themselves at E2. But for filling out a squad, three E1s is a better statline purchase than one E2.
The half-built operator is doing more than it looks
An operator sitting at E1 max already carries a median 83% of its final ATK, 83% of its final DEF and 78% of its final HP. That is a substantially complete unit. The instinct that an un-E2'd operator is "not done" overstates the gap considerably, at least in stat terms.
RES is not a universal stat
Of the 126 six-stars, only 60 have any RES at all at maximum level. Slightly more than half. Comparing operators on RES is meaningful within the classes that have it and meaningless across the roster, which is why RES rows simply do not appear on a unit page whose cohort all sits at zero.
The one thing to check yourself
These are medians across 126 operators, and individual curves vary. If you are deciding about a specific operator, the honest move is to open its page, pull the level slider to the level you would actually stop at, and read the numbers — the Stat Calculator computes from that operator's own curve rather than from any average.
The median is a description of the design's shape. It is not a substitute for the operator in front of you.