Helldivers 2: Weapon vs Enemy Calculator

Per-part breakdown

Hover a “Dmg/shot” cell for the damage maths, or a “Kill” cell for TTK, magazines and throughput detail. Shot counts assume every shot hits the same part.

◂ Prev

Next ▸

Extra columns

Shot counter Compact

Shots to kill, by part

Fewest shots aimed at one part that kill the enemy (fatal-part destruction or main-health depletion). Parts no selected weapon can kill through are omitted.

Overview

Every enemy in the current faction filter against the selected weapons, using each enemy's best part, the one that kills in the fewest shots. Click a column head to sort. With two or more weapons, the leader in each column is highlighted like this, scored per column, so the weapon that needs the fewest shots is often not the one that kills fastest.

Glossary

Every section below, collapsed by default: click a section's own title to expand it in place, or jump straight to it here.

TL;DR

The one-paragraph version of everything below: read this if you read nothing else.

• Standard ballistic weapons have the most reliable and well-modelled numbers here: direct hits, no time-dependence beyond a dropped reload.
• Continuous-fire weapons (flamethrowers, lasers) are approximated as 60 pseudo-shots per minute, should be mostly reliable, though the fuel-to-seconds conversion and buildup strength both rest on our own calibrated constants rather than published ones.
• Damage over time (fire, gas) always counts towards Kill and TTK and drains main health through main's own armor rather than the aimed part's; a burn can kill straight through a part every round bounces off. Its decay rate and application strength are calibrated assumptions, not published values.
• Arc weapons don't factor in that they chain to further targets (up to three, with a booster); the figures here are the first hit only, so they're weaker than the weapon's real, crowd-clearing performance.
• Explosives are heavily simplified: blast radius and splash onto other body parts isn't modelled, shrapnel and cluster bomblets aren't counted; figures here are weaker than reality.
• Damage fall-off is based on an unofficial curve, fit to community-tested range data: mean error 0.44 percentage points, worst case 4.06pp, so treat it as indicative rather than exact.
• Reload duration has no published source for almost any weapon and is dropped entirely; a kill needing a second magazine reads faster here than it would in-game.
• A dozen weapons have selectable states (charge levels, fire modes); the headline number is always the lowest/first one, so pick the state you actually care about on the weapon tile.
• Accuracy, stagger, overpenetration and mission modifiers are never modelled at all; every number on this page assumes a stationary target and every shot landing.

These numbers are a ceiling: a stationary enemy, every shot on the same part, every pellet landing, direct angles, no misses and no incoming fire. Use them to compare weapons against each other, not to predict how long something will take in a real fight.

Reading the tables

About

What this calculator works out exactly, what it approximates, and what it cannot see at all. Every number elsewhere in the page is produced by the rules below; if one of them doesn't hold in your game, the number won't either.

Enemy and weapon stats scraped from helldivers.wiki.gg on - · - · ignite thresholds, exploded-target flags, stagger tiers, magazine variants and multi-mode fire rates are parsed from the game's own data files (refreshable per patch); Explosion and Fire/Gas damage entries scraped from the wiki's own attack tables and refresh with the roster. Reload duration has no published source at all and is dropped; see the How shots-to-kill and TTK are found card for what that means for a multi-magazine kill.

Explosions Approximate: heavily simplified

An explosion entry hits the main pool always, and the aimed part too where the wiki marks that part explosive-targetable, each scaled by its explosive resistance. Explosive damage is treated as fully durable, and armor still applies at the blast's own penetration.

The blast is a different pool from the part, and the table keeps them apart. A part's ExDR is how much explosion damage it shrugs off, and at 100% (every Devastator part, - of the - parts in the roster) it takes none: the wiki's rule is that the blast then targets main health instead. So “Dmg/shot” is what the part takes, because that is the number “Part” (shots to destroy) divides, and the blast is shown next to it as +N → main. This is why a weapon can read 0 damage against every part of an enemy and still carry a Kill number: the SG-8P Punisher Plasma's projectile does literally nothing (0 / 0 on the wiki) and its 225 blast kills a 750 HP Devastator in 4 shots, from any part you aim at.

• Splash landing on other body parts cannot be accounted for. A real blast hits several parts at once; this model credits at most two pools.
• There is no blast radius, no inner/outer damage falloff and no line of sight: no notion of where the round actually landed, and no distance term (see above).
• Where a part is not explosion-immune and is marked explosive-targetable, the blast is credited to the part and to main. The wiki's own note says the redirect to main happens once per explosion and not if that explosion already damaged main, so this is arguably generous on those - parts. It is kept because it is what the reference dataset's own calculator does, and because it pulls the other way from the unmodelled splash below.
• Consequently explosive weapons are generally understated against multi-part enemies, and, with the stop-at-part-death rule, understated again on armor plates.
• The wiki does publish inner/outer/shockwave radii, outer-radius damage falloff and shrapnel count for - weapons' blasts (shown dimmed on the weapon tile's damage panel) but this flat single-value model deliberately ignores all of it: no radius, no falloff curve across the blast, no shockwave.

Shrapnel and cluster bomblets Not modelled

Some weapons carry a third damage entry beyond the bullet and the blast: fragments thrown out of the impact point, or bomblets scattered by an airburst. None of it is counted. - in the roster carry some: -.

The reason is that the count is a ceiling, not an amount. Where 30 fragments go depends on the geometry of the impact and the target, and nothing in the data says how many of them land on the part you aimed at. Crediting all 30 of the Eruptor's would add 3,300 damage to a shot and one-shot most of the roster; crediting one is as arbitrary as crediting none, without being any more honest. So the shot counts for these weapons are a floor: the real weapon is better, most of all at close range where more of the spray connects. The full figures are listed on the weapon tile, marked not modelled, and named in the caveats panel whenever one of these weapons is selected, so the shortfall is visible rather than silent.

Stun entries are here for a different reason: they carry no damage at all, no enemy has a stun susceptibility in the data, and nothing in the simulation reacts to being stunned.

Fire and gas Approximate: calibrated assumptions

Every shot adds status strength × pellets to a buildup meter. At the enemy's guaranteed threshold the effect ignites and ticks against main health (scaled by that enemy's fire multiplier) for as long as the meter stays above zero; further hits top it up, so keeping something burning costs shots, and the simulation counts those shots. Three things in that sentence are assumptions:

• The decay rate is calibrated, not published. No public dataset carries one, so it is set to threshold ÷ duration, the one rate that makes a bare ignition burn for exactly the effect's listed duration. That duration is per-weapon, not a constant, typically ~3 s for fire and ~6-10 s for gas (the TX-41 Sterilizer's Gas runs 10 s).
• The meter is capped at the ignite threshold, so top-ups refresh a burn but can't bank extra time. Without the cap a single 100-strength gas dart would buy roughly ten minutes of gas against a threshold-1 enemy.
• Only the guaranteed threshold is used. Susceptibility data has a lower bound above which ignition is a chance; that region is ignored, so ignition is always modelled as later than it might happen. - of - status entries publish a min ≠ max (the Status Thresholds panel already shows this as 4 (from 3)), so for most enemies the simulated ignite point is a worst case, not the earliest it could plausibly happen.

The thresholds themselves are parsed directly from the game's own StatusEffectSusceptibility data (not the wiki, which doesn't publish them) and are refreshed by rerunning the datamine after a patch. 10 enemies, mostly vehicles and structures (Annihilator Tank, Barrager Tank, Dropship, Gunship, Shredder Tank, Vox Engine, Illuminate Overship, Stingray, Warp Ship, Hive Lord), have no ignite data at all; there is no fallback for them, fire/gas simply never ignites. The gas cloud isn't modelled at all: no lingering area, no second enemy walking into it, no enemy walking out. Electricity and stun are display-only: the data carries no DoT numbers for electricity, and nothing in the model consumes a stun or a stagger.

A DoT damages main health only, and against main's armor. Not the armor of the part you aimed at: the wiki is explicit about this, and it has a large consequence: a burn kills straight through a part that every round bounces off. A Breaker Incendiary's AP2 pellets all ricochet off a Charger's AV4 head for nothing, but the fire tick is AP4 against an AV4 main, so it does 100 × 1.5 × 0.65 = 97.5 dps and the Charger dies in 13 trigger pulls. The tick is scaled by the enemy's own fire multiplier and by the same equal-armor 65% reduction as any other damage.

Fire and gas always count towards Kill and TTK. Until 2026-07-26 a three-way “DoT in kill” option defaulted to counting a burn only for weapons whose damage was the effect, a structural guess at a flag the data doesn't carry. It went for two reasons. First, excluding a lit burn produces shot counts that are simply wrong: if ballistics alone need 14 rounds but the fire already burning would finish the job after 13, you never need the 14th, and 178 rows shifted like that. Second, the case that justified the exclusion, the ricocheting Charger above, turned out to be correct game mechanics rather than a modelling artefact, so the option was hiding a real answer. One Fire/gas DoT setting remains, for stripping the burn out to check raw ballistics; the DoT is simulated and reported in the Ignite and DoT DPS columns either way, the latter marked not in TTK when excluded.

Kill is the fastest answer, not the cheapest; the Min ammo column is the other one. The sim fires as fast as the weapon allows, so a burn only ever gets the gap between two shots to work in, and one more round nearly always beats waiting for it. Where Min ammo shows a number you can stop shooting there and the burn already lit finishes the job: an R-4 Hyena kills an Agitator's pelvis in 4 rounds in 0.95 s, or 3 rounds in 2.12 s if you let the fire do the last 74 HP. It is bounded by construction: the buildup meter caps at the ignite threshold, so the tail after your last shot can never exceed the effect's own duration, which is what stops this degenerating into “fire, wait, top up, wait again”. It is blank where ceasing fire never kills: with no direct damage at all the burn needs topping up, so the Kill figure is the only answer there.

Arc weapons Approximate

- weapons, the ARC-3 Arc Thrower, ARC-12 Blitzer, and the G-31 Arc throwable, fire an arc rather than a projectile: no mass, velocity or drag exists for them, so damage fall-off isn't modelled at all, only a hard range cut-off. The two guns publish no Fire Rate or Capacity row on their stat table; both are read from the infobox instead, and their true "∞" capacity is carried as a dedicated flag rather than a magazine size, so they never reload. The G-31, being a throwable, has neither row to begin with; see the Throwables card.

• You can't aim an arc at a specific part. The wiki is explicit that the arc cannot be aimed at specific limbs: facing an enemy head-on it hits the head, from the side the body. This undercuts the premise of the entire per-part table for these two weapons: every other row on this page assumes you chose the part you're looking at, and for these two you can't.
• Chaining isn't modelled. Each arc chains to one further target (three with the Circuit Expansion booster) at 1 / (1 + chains) damage: 100% down to 50%. The figures here are the first target only, which understates a weapon whose actual job is clearing a crowd.
• Line of sight isn't modelled. An arc needs roughly a cubic metre of clearance and grounds itself into corpses, fences and walls; the initial bolt has a 3° firing angle.
• The Blitzer's 5 arcs are modelled like shotgun pellets, all five assumed to connect on the same part, matching its own wiki tactical text ("each salvo with a total damage of 250 is divided into five arcs with damage of 50"). The wiki notes the real total depends on how many arcs actually connect, so this is a best case.
• Arcs chain indiscriminately, including onto Helldivers and civilians, not something a per-enemy table can represent.
• Stun is display-only, same as elsewhere: no enemy carries a stun susceptibility, so "shots to stun" is unknowable and there's no column for it. What's known, status, strength, duration, is constant per weapon and shown on the weapon tile's damage panel instead of repeating one value down every row.

Continuous fire Approximate

- weapons, flamethrowers, lasers, the Cremator, have no discrete rounds, so they're modelled as 60 pseudo-shots per minute, one per second of sustained fire. Every shot count in the tables and tiles for these weapons is therefore seconds of fire, not rounds; the Kill column header still says “Shots” because at 60 rpm the number already reads that way.

• Damage is the wiki infobox's DPS figure, not the attack table's. A spray's attack table publishes per-particle damage (the Torcher's reads “2 Fire”) with no published particle rate, so there's no honest way to turn it into a per-shot number. The infobox's Standard Damage row gives the per-second figure directly (150 DPS for the Torcher) and that's what the simulation uses; the per-particle figure is shown on the tile for reference only.
• Magazine size, for a spray, is the wiki's fuel figure divided by our own measured constant. The wiki's Capacity row for these weapons is fuel units (the Torcher lists 100), not seconds; feeding that into the sim would claim 100 seconds of continuous fire. FUEL_PER_SEC (in scrape.py) converts it to real seconds of fire; it isn't game data, it's our own fit against the surviving measured points. Beams use the wiki's own numbers directly instead: Overheats at ÷ Heat per second gives real seconds of fire (the Scythe's 100°C ÷ 12.5°C/s = 8s, matching its infobox capacity), so no conversion constant is needed there.
• Buildup strength is the wiki's own per-particle figure, scaled by a constant we calibrated ourselves (STATUS_APPLY_HZ in scrape.py), not read from any third party. A straight rescale to the particle rate (~75/s for a flamethrower) would ignite everything in about a second, which makes the Ignite column meaningless, but the game's own data has no name-joinable source for the real application rate either (checked against the datamine's DamageSettings table, which carries a per-effect strength but no way to tell which of its ~600 anonymous rows belongs to which weapon), so the constant is calibrated against a sanity anchor instead: a dedicated flame weapon's sustained buildup rate shouldn't undercut the SG-225IE Breaker Incendiary's wiki-sourced one. That anchor is met for the strongest flame weapons at the cost of them igniting every enemy in a single second; the weaker ones (Torcher, the beam lasers) still discriminate by a couple of seconds.
• The enemy's fire/gas damage multiplier still applies only to the DoT tick, exactly as for any other weapon, not to a spray's direct damage. That's a property of the status effect in the game's own data, not of fire damage in general, and it moves the numbers by 0.5×-1.8× across the roster, so it's worth knowing it's a modelling choice rather than an oversight.
• The SMG/FLAM-34 Stoker also has a page-level flamethrower spray table, but it's a separate underbarrel fire mode from its primary SMG rounds; crediting both at once would double-count a trigger the game only lets you pull one of. It's parsed as the ordinary SMG it primarily is; the underbarrel spray isn't in the data at all.

Melee weapons Approximate

- CQC-* weapons are melee. What's exact about them is ordinary: damage, durable damage, AP per angle and rate of fire are the wiki attack table's own values, read the same way as any gun's. What isn't modelled is everything that makes melee its own thing in the actual game.

• Range is treated as 0 m with no fall-off curve. Melee weapons publish no Mass, Initial Velocity or Drag Factor, so the distance slider doesn't apply to them; the tile already reports "fall-off not modelled" for free, the same as any other weapon missing that data.
• Reach and swing arc don't exist in the model. A melee weapon "hits" whatever part you aim at, exactly like a rifle does; there's no cone, no multi-target swing, and no notion of being too far away to connect (short of the 0 m assumption above).
• Stagger and stun are display-only, same treatment as arc weapons: no enemy carries a stun susceptibility, so there's no column for it, and the strength/status shown on the tile never enters a calculation.
• The CQC-9 Defoliation Tool is a chainsaw with no published fire rate at all; it's modelled as continuous fire (see the card above), the only melee weapon that is.
• The CQC-20 Breaching Hammer also publishes no fire rate anywhere on its page; rather than being dropped, it's in with the same "no cadence" flag the throwables card below uses: shots-to-kill is exact, TTK/DPS/reload read "-".

Throwables Approximate

- grenades are in: every throwable that carries a real Damage section on its wiki attack tables. What's exact about them is the same ordinary rule as any other weapon: damage, durable damage and AP come straight from the attack table, and the blast's shrapnel/bomblet counts are read the same way an exploding rifle's are. What's approximate or missing is what makes a thrown weapon different from a fired one.

• No throw carries a cadence, so there's no TTK, DPS or reload. No grenade publishes a fire rate anywhere on its page; inventing one (a 60 rpm assumption, or reading the infobox's Fuse Time, which is detonation delay, not throw cadence) would be a fabrication. A dedicated noCadence flag suppresses those columns instead: shots-to-kill stays exact, TTK/DPS/reload read "-" with this disclosure. The CQC-20 Breaching Hammer (see the melee card) uses the same flag for the same reason.
• Capacity comes from "Max Rounds", and there is no reload. A grenade is resupplied between engagements, never reloaded mid-fight; the tile shows the real number carried, not "∞" (unlike the two arc weapons, this number is small and meaningful).
• The blast IS the damage for every throwable except the K-2 Throwing Knife (a genuine thrown melee weapon, no explosion at all) and the G-31 Arc (whose own "explosion" table is a 0-damage stub pointing at its real Arc table). Everywhere else, the tile's Damage/Durable stats read 0, same convention as the SG-8P Punisher Plasma, and the real number is the Explosion row in the damage panel.
• Shrapnel and bomblets are display-only, same treatment as any exploding projectile weapon: the G-6 Frag and TM-1 Lure Mine's 35 shrapnel pellets, and the G-7 Pineapple's 18 cluster bomblets (a ballistic hit plus their own smaller blast), are shown on the tile and excluded from the sim; how many land on the part you aimed at isn't in the data.
• The G-123 Thermite is a known partial: its blast carries no ignite trigger at all (no Status row on the explosion table), so it ships as a plain 2,000 Anti-Tank III blast. Its Fire tick entry parses and displays, but never ignites in the simulation; there's no published strength anywhere on this weapon's page to key an ignition off. Disclosed here rather than silently dropped.
• Six discovered throwables are still out, all zero-damage or uncached: G-3 Smoke and G-89 Smokescreen (radius/duration only, no Damage section at all), G-23 Stun and G-109 Urchin (0 Explosion, Stun Large only, nothing the sim can use), and the G/SH-39 Shield (a shield backpack whose wiki page isn't even cached).

How one shot is resolved Exact

Every shot in the simulation is priced the same way, in this order:

• Armor first. The weapon's penetration is compared to the part's armor value: AP > AV deals 100% damage, AP = AV deals 65%, AP < AV ricochets for nothing at all, not reduced damage, zero. A ricochet is a dead end unless something part-independent (a counted burn, a blast on the body) is also running.
• Durability split. Each part has a durability percentage, and the shot pays the weapon's standard damage on the non-durable share and its durable damage on the rest: standard × (1 − durability) + durable × durability. This is why a high-damage weapon can lose to a weaker one on a 100%-durable part.
• Pellets. Multi-pellet weapons list damage per pellet; the shot is multiplied by the pellet count, and every pellet is assumed to hit the same part. Real buckshot spreads, so shotgun figures are a best case.
• Fall-off multiplies the result when the distance box is non-zero (see below).
• Transfer to main. Damage into a part also drains the enemy's main pool at the part's “→ Main” percentage. Whether overkill carries through is the anatomy table's “Overflow Cap?” column, and the table shows it: a percentage marked ovf is uncapped, and every point of damage transfers however far past the part's own health it goes. Without the marker the transfer is capped at the part's combined health and constitution; you can't carry more through the part than the part had. The wiki's own example: a Trooper's arms are 65 HP at 50% to main and uncapped, so a single 300-damage R-6 Deadeye round puts 150 into a 125-HP body and kills outright, while the same round into a capped part would transfer at most 32. Just over 200 of 525 parts are uncapped, so it is the common case, not an exotic one. Where the column says nothing (48 parts, including every Conflagration Devastator part) the cap is assumed.
• Constitution and bleed. A second pool behind a part's health. Where the part bleeds, it drains per second once health is gone (and “Kill means: fully destroyed” waits for it); where it doesn't, it is simply extra health, shown as 1,500+2,000.

The same arithmetic prices the main pool, which carries its own armor value and durability: that is what a fire tick or an explosion is charged against when it reaches the body.

How shots-to-kill and TTK are found Exact

Not a formula: a per-shot simulation, because damage over time, bleed and reloads make the timeline path-dependent. The loop runs at the weapon's trigger cadence and:

• only fires when a trigger pull achieves something: direct damage, progress towards ignition, or topping up a burn that is about to run dry and still has work to do. Idle ticks just let time pass. So every counted shot did real work; there are no phantom shots inflating a total.
• doesn't wait for a reload at magazine boundaries: nothing publishes a reload duration for almost any weapon (not the wiki, and the game's own data is 0 or missing for most of the roster), so rather than estimate one the calculator adds none. A kill needing more than one magazine shows a +N reload chip alongside Kill/TTK as an explicit disclosure that the real time is longer. The 4 weapons whose one-round cycle time is published (a charge/warmup/cooldown weapon's whole cycle, e.g. the Railgun) are the exception: that figure is derived from the wiki's own effective fire rate, not a reload estimate.
• starts the clock at the first shot, which lands instantly at t = 0. Aim time, trigger discipline, first-shot recoil and reaction time are not in TTK, so a 1-shot kill reads 0.00 s.
• stops when the enemy dies, fatal part destroyed, or main health depleted; “Kill means” chooses whether a bleed-out counts as dead. A kill finished by a DoT or bleed after the last shot shows the tail, e.g. 6 +0.7s burn.

With no DoT, no constitution and no reload the simulation reproduces the closed-form arithmetic exactly; that property is regression-tested.

Damage fall-off Approximate: our fit to unofficial measurements

The wiki describes fall-off only qualitatively. The curve here decays projectile velocity by 1 m numeric integration (deceleration proportional to calibre² ÷ mass, textbook drag physics) and retains damage as (v/v₀)^E, floored at 25.8% retained (i.e. loss never exceeds 74.2%); the three constants (K, β, E) are our own fit to "Helldivers II Index - 6.3.0 [UNOFFICIAL] - Falloff.csv", a community-compiled, manually-tested (~±0.3%) range-testing dataset covering 77 weapons × 8 distances, 602 measured points across the 76 rows the fit is validated against (the GP-20 Ultimatum has its own override and is excluded). Mean error 0.44 pp, worst 4.06 pp, better than the five-constant third-party fit this replaced (0.51 pp mean / 6.33 pp worst). It is our fit now, but the underlying measurements are still unofficial, not published by the game or the wiki.

• Shotguns are among the best-validated rows in the dataset: every shotgun fits within 0.3 pp, and an SG-225 Breaker keeps 76% of its damage at 50 m. (An earlier version of this page claimed shotgun fall-off was unreliable; the measurements say the opposite.)
• Loss is clamped at 74.2%, an empirical floor read off the measurements: five plasma-weapon rows land on exactly 74.2% at 200 m where the curve alone predicts more, and the CSV's own header calls 74.2% a peak. Not a claimed game mechanic.
• It needs velocity, mass, calibre and drag; several weapons don't list all four and simply never fall off. Their tiles say so. 7 in-roster weapons (FAF-14 Spear, MLS-4X Commando, GL-21 Grenade Launcher, GL-28 Belt-Fed Grenade Launcher, R-4 Hyena, SG-8P Punisher Plasma, StA-X3 W.A.S.P.) are absent from the measured dataset; they still fall off, but as untested extrapolation of the fitted curve.
• Fall-off follows the selected weapon state for mass, velocity, drag and calibre; each state carries its own projectile stats from the datamine, so e.g. the SG-20 Halt's two states (8 mm and 5.5 mm) fall off on genuinely different curves, not a shared one.
• The GP-20 Ultimatum is the one weapon the formula cannot express: its slow round gains damage with distance as it falls under gravity. It uses a measured lookup table instead of the curve; its tile says so.
• Three CSV rows measure a different weapon than their label and are excluded from validation: the PLAS-101 Purifier and PLAS-15 Loyalist rows both carry the PLAS-1 Scorcher's projectile data, and the RL-77 Airburst row carries the GR-8 Recoilless Rifle's. Our numbers for those three rest on the wiki's projectile stats and the fitted curve, with no direct measurement behind them. Two more weapons, the AR/GL-21 (calibre 40 vs the sheet's 8) and P-4 Senator (13 vs 12), disagree with the sheet on calibre and we have not resolved which figure is right.
• Only direct damage falls off. Explosion entries are applied at full value at any distance, so a blast weapon's advantage grows artificially with the distance box.
• Distance is the only spatial input (capped at 200 m, the measurements' limit): no angle of attack, no drop, no travel time, no target movement. That's also why the Ultimatum needs its own override rather than a gravity term: modelling it honestly needs trajectory, aim angle and drop, none of which exist here.

Weapon states Approximate

- weapons have more than one damage/AP figure depending on a state the player controls in-game: a charge held longer (the RS-422 Railgun, PLAS-39 Accelerator Rifle, and the PLAS-15/45/101 Loyalist/Epoch/Purifier), a heat ramp that never locks out (the LAS-17 Double-Edge Sickle), or a fire-mode switch (the AC-8 Autocannon's APHET/FLAK, and five others). The weapon tile gets a state selector next to its fire-rate/magazine picks, and the currently selected state is always named on the tile and in the caveats panel; showing one unlabelled number per weapon would have been worse than not modelling this at all, since three of these weapons used to show three different, incomparable charge states without saying so.

• Each state's figures are the wiki's own published block for that state, never an interpolation. A charge weapon's table publishes exactly the states that get their own selectable step (a single-block charge weapon's table samples three points along a ramp, but the middle one is a sample, not a level; only the first and last become states, with the multiplier applied to damage only, never to AP).
• Charge time replaces the fire interval, not adds to it. A charge weapon's cadence is chargeTime itself, not 60/rpm plus the hold; the wiki's own effective rate for these weapons decomposes as charge time + reload, and treating the two as additive would double-count the cycle. The default state is always the lowest/first one, which moved the headline number for two weapons: the PLAS-101 Purifier used to show its fully-charged figure and the RS-422 Railgun's overcharge-only blast used to apply to every shot, including safe ones.
• For the LAS-17 Double-Edge Sickle specifically: its four heat stages are ordered but carry no published temperature threshold; the wiki states the weapon overheats at 200°C and never says how much heat buys which of the four projectile variants, only that stage 2 trades some damage for more penetration on the way up. Labelled as stages, never as a temperature.

Shots stop at the aimed part Modelled: routed by default

Every row's own per-shot simulation still assumes you keep hitting that one part: once it is fully destroyed, that sim stops firing there; it never guesses which part you'd switch to. What changed is the Kill column itself: for - parts across - enemies the “part X exposes part Y” chain is documented (hand-curated, because the game's own data doesn't publish this relation; see the per-part table's breaks → / behind chips), and Kill now routes through that armour by default: it counts the shots to destroy the cover, then the shots into the part behind it, on the same clock as everything else. See the Kills through armour card below for the detail. The route is a sequence of two per-row sims chained together, not a new kind of simulation, which is exactly why the per-row sim still has to stop at the aimed part: that's the building block the route is made of. The exposure map itself is still the one field on this page that's editorial rather than scraped or datamined; a human read the enemy's Tactical Information and the game's own zone-visibility flags and decided, the way a wiki citation or a datamine join can't.

A row reading “part only” or “can't kill” still means what it always did: destroying that part alone doesn't kill anything, whether or not it's covered. Where a part is covered, its Kill cell shows the routed total instead, a struck-through number when this weapon can't break the cover, or the same unmodified “part only”/“can't kill” where the cover is breakable but nothing behind it kills either way. The one thing routing still can't show is a route through a second weapon; the Shot counter is for that.

The alternative was worse: without the stop-at-death rule, anything with a part-independent effect ground on against wreckage forever (a gas weapon “killing” through a torso plate that no longer existed by re-gassing it).

Kills through armour Approximate: one weapon, one route

The Kill route option (in the per-part table's option bar, “Through armor” by default) is what actually does the routing described above. It counts the shots to destroy a covering part, then the shots into the part behind it, replayed on the Shot counter's own shared-pool engine, never a second, disconnected estimate, and never inside calc() itself (that function stays direct-hits-only; routing lives entirely in the wrapper around it).

• Three cell shapes. A reachable route shows the total with a quiet 4 → Helmet line underneath (hover it to highlight the cover row). A route that exists in principle but this weapon can't break the cover shows the part's own direct number struck through, with a can't break … chip; the number is real information (“this weapon would kill through here if something else opened it”), it just isn't a plan on its own. Where the cover is breakable but nothing behind it kills either way, including when this weapon can't damage the covered part at all regardless of the cover, e.g. its penetration is too low for the part's own armor, the cell reads the same “part only”/“can't kill” as an uncovered part, with no chip: the cover isn't what's stopping the kill, so there's nothing to blame on it.
• A covered part is unreachable from every angle until its cover is fully destroyed: health and constitution both gone. The game is more forgiving than that in places: the Scout Strider's Pilot is shootable from the side in-game, but this model makes you break the Front Plate first. That's the one assumption here that's pessimistic rather than optimistic, unlike most of this page's simplifications.
• The route is one weapon, shot in sequence. Breaking the armour with a second weapon and finishing with a first is exactly what the Shot counter is for; the Kill column doesn't guess at loadout tactics.
• Only - of - enemies have a curated exposure map at all (- covered parts total). An enemy that isn't mapped, or a part on a mapped enemy that isn't covered, still shows its unrouted, optimistic direct number; that's the exposure map's own coverage gap, not a new one this task introduces.
• Some covers hide a part the wiki only documents after the cover breaks. - such parts across - enemies come straight from the wiki's own post-armour-break anatomy table, a real, separate row (Internals (2) behind Harvester's Carapace (2), for one), not a stat-swap of the intact part. They're collapsed by default behind a ▸ next to their cover's name, a different glyph from the Shot counter's own +/- buttons, so the two never collide on the same row, and auto-reveal the moment the counter shows that cover destroyed. The honest limit: only 4 of 83 enemies publish such a table at all, so every other enemy's post-break anatomy simply isn't documented anywhere, on the wiki or off it.

Switch Kill route to “Direct hits only” to see the old, unrouted numbers: what the part itself takes, ignoring what's in front of it; the whole page (per-part table and Breakpoints both) renders exactly as it did before this feature existed. Only Kill, TTK and Min ammo move between the two settings; Min ammo is blank on a routed row, because the ammo-optimal count for the part alone isn't the ammo-optimal count for the whole route.

Shot counter Approximate

The Shot counter toggle (in the per-part table's option bar, off by default) turns the table from "shots at one part, one row at a time" into a real, ordered sequence: click a row's + to fire that weapon at that part, click again for another part, and watch a single shared pool of enemy health drain as you go, the payoff of the hand-curated exposure map above, since the model never has to guess which part you'd aim at next once one breaks. It is a real timeline, not instant damage accounting: each shot fires at that weapon's cadence, in click order, with fire/gas buildup, DoT ticks, bleed-out and reloads running in the gaps exactly as the ordinary per-row simulation does, the same shotProfile() arithmetic, shared verbatim, so the two can't quietly disagree. A part with an undestroyed cover shows a disabled + naming what to destroy first (hover it to see the cover row highlighted); an already-dead part or an already-dead enemy disables it too, with the reason in the tooltip.

The Kill column's own automatic routing (see Kills through armour above) now covers the single-weapon case this used to be the only way to see: one weapon, cover then target. What the Shot counter is still for is everything routing can't guess: a second weapon finishing what the first opened, hitting parts in an order the router wouldn't have picked, or just watching the shared pool drain shot by shot. Here you sequence the shots; the model only plays out what you actually clicked.

Two things it deliberately simplifies, stated here and in the summary strip above the table:

• Switching weapon or target costs no time. There's no published number for how long a reposition or a weapon swap takes, and inventing one would be worse than declaring the assumption, same reasoning as the dropped reload duration elsewhere on this page.
• One buildup meter per status name, shared between every weapon that can apply it: a flamethrower and an incendiary sidearm feed the same fire, not two independent burns. The meter carries whichever weapon's numbers (threshold, decay, DPS) applied it most recently, which is exact when every weapon publishes the same figures for a status and an approximation when they don't.

The ledger itself (which shots you fired, in what order) is session state, not shareable: it isn't in the URL hash and isn't kept between visits, only the toggle's on/off preference is. Undo removes the most recent shot; Reset clears the whole sequence.

The clock stops at your last shot. A lit burn or an open bleed isn't run to its natural end and folded into the reported state: that would credit you with damage and a kill you haven't actually earned yet, and lock the + buttons behind it. Instead the summary strip reports exactly where things stand the moment you stopped clicking, and shows the burn's natural finish as a separate, clearly-labelled projection: "will die to fire in 1.9s if you stop shooting" rather than a silent "killed." Keep clicking and, if the enemy is still alive, a real kill lands on whichever shot actually does it. This is the counter walking the table's own Kill and Min ammo columns by hand: Kill is what you get firing continuously to the end; Min ammo is what you get stopping at the point the counter's own projection would call fatal.

Main body row

The per-part table's first row, Main body, is the enemy's own health pool, the same pool the wiki's anatomy table lists as Main, and the same one every → Main transfer, explosion and DoT tick elsewhere on the page already drains. It is not an anatomy part, and it isn't stored as one: it's built on the fly from the same enemy.main figures the tile's "Main HP" stat already shows, with no engine change; it's simulated by calling the same per-shot loop with a pseudo-part whose transfer is 1:1 and uncapped, so the result is exactly the enemy's main pool being shot directly, not a second, parallel calculation.

It exists because an enemy whose every anatomy part is either non-penetrable or merely breakable used to read as unkillable, even when the body itself is comfortably killable: the Hive Guard's head is Medium armor and every other part transfers 25-45%, so every part row says "part only" against an AR-23 Liberator, but its 500 HP Light-armor main pool dies in 12 shots. → Main and Break don't apply to a pool that is the body, so both read “-”, and a losing row reads “can't kill” rather than “part only”, that phrase specifically means no part transfers enough, which isn't true of this row.

It is not a free pass: a Heavy-armor main still ricochets a light-pen weapon exactly like a Heavy-armor part would (the AR-23 Liberator can't kill 32 of 83 enemies through their main body). And it assumes you can hit what you aim at: on many enemies the body is only exposed from certain angles, or only once armor plating is stripped away, which this row has no way to represent. Treat a main-body-only kill as optimistic in the same way an unarmored-flank shot is: mechanically real, situationally conditional.

The wiki's Main row means two different things, though, and this page tells them apart. Some enemies have a real, shootable surface that belongs to no anatomy part; the wiki's photo highlights it (Hive Guard's abdomen, Harvester's small torso patch). Others have no such surface at all: every visible part of the body belongs to some anatomy part, and "Main" is only the pool that transfer damage, blasts and DoT drain; the wiki's photo shows this as a generic green "MAIN" label over an unhighlighted body (Overseer, Devastator, every tank). scraper/classify_main.py tells the two apart by template-matching the green pixels in each Main-view photo against that fixed label; - enemies have a real aim point and - don't. An enemy with no aim point still shows the row; it's still what every transfer/blast/DoT drains, and occasionally (Leviathan) the only kill available, but it carries a "no aim point" chip, reads dimmed, and is never picked as the Overview tab's "best part to shoot": there's nowhere to point a weapon at it.

Repeated parts Approximate

A part named Upper Arms (2) is two arms, and the wiki's health/armor figures are per arm. When one is destroyed the simulation moves to the next identical instance and keeps firing, up to the published count; unlike the “what's behind a part” case above, the next thing you shoot is in the data. - of - parts carry a count above 1, across - enemies.

• No repositioning time is charged. Moving from one arm to the other is free here; a tank's Tracks (2) are on opposite sides and really are not. TTK is optimistic by however long the move takes.
• No reachability check. Whether all 12 of a Bile Titan's leg plates can be hit from where you're standing isn't modelled, same assumption the rest of the page already makes about hitting the part you aimed at.
• Break is still one of them. Shots-to-destroy is per instance; only Kill spans the set, and it says so with a ×N marker.
• Destroying every instance is still not a kill condition in its own right; a Hulk dying to both legs is the separate multi-part case above, and stays unmodelled.

Kill count Approximate: one full load, ignores resupply

The Overview's Kill count column is a weapon's own Capacity plus its Spare Magazines / Spare Rounds / Spare Shells figure (whichever the wiki publishes, times the selected magazine variant), divided by that row's Kill shot count: the total number of that enemy the weapon can kill using all available ammo, including spare mags. It exists because a weapon that looks strictly better shot-for-shot can still lose badly on ammo: the AR-23 Liberator kills a Devastator's head in 2 shots against the R-63CS Diligence Counter Sniper's 1, but the Liberator carries 405 rounds to the Diligence's 105, roughly 202 kills a load against 105.

• It ignores resupply. This is one load off spawn, not a mission's worth of ammo boxes and Hellpods.
• Infinite-ammo weapons (melee, the two arc guns, the LAS-17 Double-Edge Sickle) read ∞ and win the column on every row they can kill at all: that's a real property of those weapons, not a display bug.
• Throwables use Capacity as the total; no spare figure exists for them, and the capacity is what you carry.
• - weapons publish no spare-ammo figure at all and read - rather than a guess: three Expendable Anti-Tanks, the LAS-99 Quasar Cannon and a handful of other recharge/belt-fed weapons among them.
• One figure worth distrusting rather than us correcting it: the wiki gives the AC-8 Autocannon a Capacity of 10 and Spare Rounds of 10, which reads low against its backpack; we publish the wiki's number as-is.

What isn't modelled at all Not modelled

• Overpenetration and cleave. A round that comfortably out-penetrates a hitbox passes through it in game; here every shot stops at the part you aimed at.
• Multi-part kill conditions, e.g. a Hulk dying to both legs. One part at a time, always. Nor is there a downed-vs-killed distinction.
• Angle-dependent armor. Parts whose armor changes with the angle of attack (“Light, Medium at Extreme”) use their direct-angle class, noted next to the part name. Every hit is assumed to be a direct-angle hit.
• Accuracy of any kind. Every shot hits, every shot hits the same part, every pellet hits. No spread, recoil, sway or movement; the recoil and ergonomics figures on the weapon tiles are shown for reference and never enter a calculation.
• Stagger and stun interactions. Enemy stagger tiers and weapon stagger force are in the data and displayed, but nothing in the simulation reacts to them; an enemy is a stationary damage sponge that never flinches, retreats, heals or attacks back.
• Two underbarrel fire modes with no selectable state. The AR/GL-21 One-Two's grenade underbarrel and the SMG/FLAM-34 Stoker's flame underbarrel are each a second, separate trigger the wiki documents on the same page as the primary, but unlike the 12 weapons with a real state selector (Autocannon APHET/FLAK, charged vs uncharged plasma, Railgun safe/unsafe, and others; see the Weapon states card above), these two are suppressed rather than modelled (NON_SELECTABLE_BLOCKS in scrape.py), because crediting both trigger pulls at once would double-count a mode the game only lets you fire one of at a time.
• Weapon customisation and attachments, which in game change rate of fire, magazine size and handling.
• Difficulty, mission modifiers, boosters, armor passives and stratagem support. Nothing external scales these numbers.
• Sub-strains, buildings, objectives and SEAF units as targets: the roster is the standard enemy list only.

Where the numbers come from Mixed provenance

• The wiki, scraped deterministically: enemy anatomy (health, armor, durability, → main, constitution, bleed, explosive resistance, fire multiplier) and weapon stats (damage, durable, penetration, rate of fire, magazine, pellets, projectile physics, stagger, recoil, ergonomics), including a second infobox read for multi-mode fire rates and whole-cycle effective rates (4 charge/warmup/cooldown weapons) that the stat table alone doesn't carry. This refreshes.
• The game's own data files, for enemy fire/gas ignite thresholds, explosion targeting, stagger tiers, and weapon magazine-capacity variants (the wiki publishes one capacity per weapon; the customization wheel's other choices, Short/Extended/Drum and similar, come from the game's own weapon-customization data instead), parsed via a filediver-based extractor, refreshable per patch independently of the wiki scrape. Wins over the wiki only for four projectile fields and explosion targeting, which the wiki doesn't publish at all; magazine variants have no wiki equivalent to compare against either.
• Reload duration has no published source anywhere: not the wiki (it lists none), not the game's own data (0 or missing for most of the roster, and nothing anywhere publishes the tactical/partial figure), so it's dropped rather than estimated. A kill needing more than one magazine shows a +N reload disclosure instead of a fabricated number.
• The stats were extracted in one pass and are not hand-verified. Most spot checks passed; a handful of entries (a couple of “destruction is fatal” flags, some missing → main percentages) are known to be worth a second look. One wrong stat quietly poisons every number downstream of it.
• The roster now covers every attack shape the wiki tables support: projectile, spray, beam, arc, melee and, as of this task, throwables. - grenades and the melee CQC-20 Breaching Hammer (rescued via the same "no cadence" flag) are in; see the dedicated cards above. What's left, all deliberate: six zero-damage or uncached throwables (G-3 Smoke, G-89 Smokescreen, G-23 Stun, G-109 Urchin, the G/SH-39 Shield), the MS-11 Solo Silo (an emplacement, explosion-only), the B/MD C4 Pack (equipment, no weapon stat table), and the P-11 Stim Pistol (correctly excluded, it heals teammates, it doesn't damage anything). That's a parser gap for none of them any more; if a weapon isn't in the picker, it's one of the eight named here.
• Flamethrowers, lasers and the Cremator (spray and beam attack tables) parse as continuous fire; see the dedicated card above for how that's modelled.

Attribution and licence

This is an unofficial fan tool. It is not affiliated with, endorsed by, or sponsored by Arrowhead Game Studios, and claims no ownership over Helldivers 2 or its assets.

• Enemy anatomy, weapon stats, Tactical Information text, and every Explosion and Fire/Gas damage entry are scraped from helldivers.wiki.gg, published under CC BY-NC-SA 4.0 (Attribution-NonCommercial-ShareAlike; pages predating February 2024 are adapted from the Helldivers Fandom wiki under CC-BY-SA). This page inherits those terms: it is offered under the same licence, and is never monetised while it carries this data.
• Enemy fire/gas ignite thresholds, explosion targeting, stagger tiers, weapon fire-rate/magazine-variant linkage and projectile physics are datamined from the game's own data files via xypwn/filediver, at a pinned commit of that project recorded alongside each generated section.
• Our fitted damage fall-off curve is calibrated against range-testing measurements from the community-compiled "Helldivers II Index - 6.3.0 [UNOFFICIAL]" spreadsheet.
• This page is an adaptation, not a copy: the wiki's published stats have been reorganised, merged with datamined values, and used as the input to simulations and curve fits. Nearly every number shown here is computed on this page, not quoted from a source. The enemy anatomy and weapon photographs are the wiki's own uploaded files, carried under the same licence as the text.
• This page's own code is copyright 2026 Undergroundpanda, under the MIT licence. The data and imagery it ships with are not MIT, and are not ours to relicense: they stay under the terms above.