Kenshi Stat Training Calculator
Work out how long a stretch of Kenshi training really takes. The skill bar is not bugged past 90 - the last ten levels are 91.82% of the entire climb to 100. Enter a measured speed and get in-game days and real minutes at 1x, 2x or 5x.
Training time Kenshi 1.0.68
Nine tenths of the grind is in the last ten levels
Every skill slows down in exactly the same way, so this picks which skill the tables below describe. It does not change the time.
The race of the character you timed. It divides back out of your figure, so it never multiplies your answer — it picks which row the race table marks.
The game reports two decimals, and 40.03 counts.
100 is the ceiling. Nothing trains past it.
Note your level, let one in-game hour pass — two real minutes at 1× — and note it again. The figure here is a measured Strength run; change it and every number on the page follows.
8 h 32 min
0.36 in-game days
3 min 25 s
17 min 4 s at 1×, 8 min 32 s at 2×
0.81%
43 d 8 h 34 min
and to 90: 2 d 22 h
Everything below level 90 is 8.18% of the whole climb to 100. The last ten levels are the other 91.82%. Your own stretch, 40 to 60, is 0.81% of it.
The ten levels after 60 cost 8 h 23 min at this speed — 1 times the stretch you just priced.
Field Medic and Robotics are the exception: they level on time spent rather than on actions, and they are the only two skills that can pass 101. Read the figures above as the shape of the curve, not as a stopwatch.
Where the grind actually is
| Levels | At your speed | Share of 0–100 | × the first ten |
|---|---|---|---|
| 0–10 | 1 h 10 min | 0.11% | ×1.00 |
| 10–20 | 1 h 27 min | 0.14% | ×1.25 |
| 20–30 | 1 h 51 min | 0.18% | ×1.60 |
| 30–40 | 2 h 28 min | 0.23% | ×2.12 |
| 40–50 | 3 h 26 min | 0.32% | ×2.95 |
| 50–60 | 5 h 6 min | 0.48% | ×4.40 |
| 60–70 | 8 h 23 min | 0.79% | ×7.23 |
| 70–80 | 16 h 23 min | 1.55% | ×14.12 |
| 80–90 | 1 d 22 h 10 min | 4.37% | ×39.79 |
| 90–100 | 40 d 9 h 37 min | 91.82% | ×835.55 |
The same climb, ten levels at a time. The shares and the multiples are fixed by the game's XP curve and do not move when you change the speed — only the times do.
Fighting someone stronger
| Opponent's skill against yours | XP multiplier |
|---|---|
| 50 or more above you | ×6.00 |
| 40 above you | ×5.00 |
| 30 above you | ×4.00 |
| 20 above you | ×3.00 |
| 10 above you | ×2.00 |
| 5 above you | ×1.50 |
| Same level as you | ×1.00 |
| 5 below you | ×0.80 |
| 10 below you | ×0.60 |
| 20 below you | ×0.20 |
| 22.5 or more below you | ×0.10 |
Only eleven skills use this: melee attack and defence, martial arts, dodge, dexterity, and the six weapon skills. Being outnumbered adds 10% per extra enemy up to 1.667× at one against eight, so the ceiling is 6.00 × 1.667 = 10× — the maximum the wiki states. Its own per-enemy ladder reaches 1.7× there; the cap is the figure consistent with that ceiling, so the cap is what this table uses.
Which race trains this skill fastest
| Race | XP multiplier | Your stretch |
|---|---|---|
| Strength | ||
| Greenlander | ×1.00 | 8 h 32 min |
| Scorchlander | ×0.90 | 9 h 29 min |
| Shek | ×1.10 | 7 h 45 min |
| Hive Prince | ×0.80 | 10 h 40 min |
| Hive Soldier Drone | ×1.00 | 8 h 32 min |
| Hive Worker Drone | ×0.80 | 10 h 40 min |
| P4 Unit | ×1.00 | 8 h 32 min |
| Skeleton | ×1.00 | 8 h 32 min |
| Soldierbot | ×1.00 | 8 h 32 min |
The multipliers are per skill. Nine of the 32 skills have no racial modifier at all, and there the nine rows are identical — that is the correct answer, not a gap. Soldierbots carry a 0.80 Martial Arts multiplier that the wiki says does not actually reduce the XP they gain; the table reports the game's own figure.
Training rates you control
| What you are doing | XP rate |
|---|---|
| Strength — a weapon needing 20+ more Strength than you have | 110% |
| Strength — martial arts at 100% encumbrance | 100% |
| Strength — walking at 70%+ encumbrance, carrying a body | 50% |
| Strength — walking overloaded, carrying nothing | 25% |
| Strength — a weapon you already meet the requirement for | 10% |
| Athletics — running at 0% encumbrance | 150% |
| Dodge — 1% less for every 1% of encumbrance | 100% down to 40% |
| Dexterity — katana (all cut, no blunt) | 100% |
| Dexterity — rusted junk polearm | 79% |
| Dexterity — jitte (all blunt) | 0% |
Rates are a share of the skill's own base XP, so the top Strength row is eleven times the bottom one — the wiki's own sentence, reproduced by the ladder. Dexterity is cut ÷ (cut + blunt) of the weapon you are swinging. The Dodge penalty stops at 60% encumbrance; a second wiki page implies no floor at all, and this table uses the one that names a number.
The arithmetic, step by step
You measured 3.3 levels in one in-game hour, starting at level 40.
Climbing 40 to 43.3 takes 9.56 XP, where 1 XP is what one level costs at level 0. So you earn 9.56 XP an in-game hour.
Climbing 40 to 60 takes 81.58 XP.
81.58 ÷ 9.56 = 8.53 in-game hours.
8.53 × 2 real minutes ÷ 5 = 3 min 25 s.
Kenshi 1.0.68 · wiki data checked August 2026
The curve, the racial multipliers and the rate ladders come from the Kenshi wiki's Guide to Training Statistics, Stronger Opponent Logic and Statistics pages. The training speed is the one figure you measure yourself; everything built on it is exact.
Kenshi stat training calculator. How long a stretch of skill levels takes, in game time and in real time.
A Kenshi stat training calculator turns one measured hour of training into the game time and the real time a stretch of skill levels costs. It also reports how much of the entire climb to 100 that stretch is, and Kenshi's curve puts 91.82% of the climb in the last ten levels.
The number Kenshi never shows you
Kenshi prints a percentage bar under every skill and nothing else. No experience total, no rate, no estimate of what is left. Behind that bar, every scrap of experience a character earns is multiplied by (1 − level ÷ 101), twice over, so one level's cost rises with the square of how close the skill sits to 101. Price the whole climb in a single unit — what one level costs at level 0 — and 0 to 100 comes to 10,100 of them, with 91.82% of that total sitting between 90 and 100. Everything below 90 is the other 8.18%. Both shares are fixed by the curve, and no input on this page moves them.
The one number the game keeps to itself is how much experience a given action awards. Those values live in the FCS, the editor that ships with Kenshi, and no public page publishes the table — so this calculator asks you to measure rather than guess. Note a skill's level, let one in-game hour run, note it again. At the served default — Strength on a Greenlander, 3.3 levels in that hour starting from level 40 — the answer is 8 h 32 min of game time and 3 min 25 s of yours at 5× speed. That default reproduces a published run: the YouTube guide titled "The Best Kenshi Strength Training Guide. 40.03-60 Str in 8hrs 32mins gametime" timed the same stretch and landed on the same figure.
Nothing else in the Kenshi space converts the curve into a time. The wiki's Guide to Training Statistics documents the mechanic and the rate ladders and stops there. The Steam guide "Efficient Character Training Guide for Kenshi Combat Classes" is arranged by combat class and carries no hours, no formula, and no mention of the 90s. Lilura1's Strength page prints the comparison everyone quotes — "the maximum Strength XP Rate Combat achievable is 100% whereas the maximum Strength XP rate Walking achievable is 50%" — plus one aggregate guess, "it can take about 32 hours of play to get a five-person squad to 70 Strength". The combat guide on gameplay.tips says "Beyond 80+ the returns really slow down" with no number attached. My Kenshi Training Guide from Zero to Hero, on the GOG forum, is the most numerate of the lot: "It takes about 250 game days to do the whole thing, I'd guess about 15 - 20 hours of play." Two hundred and fifty in-game days is 200 real hours at 1× and 40 at 5×, and that conversion is the piece none of them prints.
PinDrop is the one site carrying real Kenshi tools: a tech-tree planner that prices a research path in Books and AI Cores, an interactive map of 304 locations, and a race comparison across 43 attributes pulled from the wiki at v1.0.68. It has the racial multipliers. It has no training calculator, and nothing on it turns a multiplier into an hour.
Most of the searching around this topic comes from a player convinced the game has broken. Steam's "SKILL PROGRESS // Why does it get stuck on 90+ ?" has a smith frozen at 92 "for over 100 in-game days" of continuous forging, and a moderator in the same thread names "an unusual bug where skills increase faster while in an NPC town". The thread "How do you get combat skills past ~70?" carries "getting it to 60 is fast, getting it to 90 really really slow" and, further down, "past 87ish you need cheese to progress". The curve accounts for most of that before any bug does. One level at 95 costs 256 times what a level at 5 cost, so a bar can be filling on schedule and still look frozen for a real hour.
Why one level at 95 costs 256 levels at 5
The experience a level costs never changes: the wiki's Guide to Training Statistics puts it as "the amount of experience needed to level up in Kenshi does not change depending on the level, instead it is based off your experience total for that skill". What changes is the experience an action pays out. Every award is scaled by the square of (1 − level ÷ 101) — 0.90 at level 5, 0.36 at level 40, 0.0035 at level 95, and exactly zero at 101. That zero is the entire reason nothing trains past 100.
Turn the multiplier upside down and one level's cost reads
so any two levels compare as a ratio of squares. Level 95 against level 5 is (96 ÷ 6)² = 256. Level 99 against level 0 is 101² ÷ 4, about 2,550. The last single level, 99 to 100, is 50.50% of the whole climb from 0 to 100 — one level that costs more than the other ninety-nine together.
Stronger Opponent Logic multiplies whatever the curve leaves you. Fighting above your own skill pays ×1.50 at 5 levels above you, ×2.00 at 10, ×4.00 at 30 and ×6.00 at 50 or more; an opponent 5 levels below pays ×0.80, and one 22.5 or more below pays ×0.10. Being outnumbered adds 10% per extra enemy up to ×1.667 at one against eight, so the ceiling is 6.00 × 1.667 = ×10, which is the maximum the wiki states. Only eleven skills read the ladder at all: Melee Attack, Melee Defence, Martial Arts, Dodge, Dexterity and the weapon skills. Steam's "Too fast to go from combat skill 1 to ~40, too slow for 40+ ?" thread shows the bottom of it in practice — a moderator points out that Leviathans sit around 40 combat stats, so a character past that is farming them at a penalty sliding toward ×0.10, which is why ten kills buy a handful of levels.
Racial multipliers are per skill and per race, and they are gentler than players expect. For Strength the fastest row is Shek at ×1.10 and the slowest is Hive Prince and Hive Worker Drone at ×0.80, with Greenlander, Hive Soldier Drone, P4 Unit, Skeleton and Soldierbot at ×1.00 and Scorchlander at ×0.90. Nine of the 32 skills — Swimming, Melee Defence, Assassination, Crossbow Smith, Crossbows, Blunt, Hackers, Polearms and Sabres — carry no racial modifier at all, so nine identical rows in that table is the right answer and not missing data. The race picker never multiplies your result. Your measured rate was measured on that race, so the engine divides the multiplier back out to recover a race-neutral base rate; changing the picker moves which row the table marks and re-prices the other rows around your figure.
Past 90 the rounding starts to bite. An award small enough rounds down to nothing and pays zero, which is why the wiki reports a hard ceiling of 96.66 Strength for a Shek training by walking with no town loaded, and why that NPC-town quirk reads to players like a fix rather than a bug. Field Medic and Robotics sit outside all of it: they level on time spent rather than on actions, and they are the only two skills that can pass 101. A Steam moderator states it flatly — "Stats that are not Field Medic or Robotics cannot naturally increased past 100, those two are the only exceptions, and do not have negative consequences for going over". For those two, read the figures here as the shape of the curve rather than as a stopwatch.
How much of the climb to 100 is behind you at each level
| Skill level reached | Share of the 0–100 climb behind you | Share still ahead |
|---|---|---|
| 10 | 0.11% | 99.89% |
| 30 | 0.42% | 99.58% |
| 50 | 0.98% | 99.02% |
| 70 | 2.26% | 97.74% |
| 80 | 3.81% | 96.19% |
| 90 | 8.18% | 91.82% |
| 95 | 15.83% | 84.17% |
| 99 | 49.50% | 50.50% |
| 100 | 100.00% | 0.00% |
Taking the one measurement the page needs
Everything on the page hangs off a single figure you time in game. Fill the form top to bottom.
1. Pick the Skill you are training. Every skill slows down in exactly the same way, so this choice drives the race table and the two note lines rather than the answer.
2. Set Race you measured on to the race of the character you timed, not the race you wish you had rolled. It divides back out of your figure.
3. Read Current level and Target level off the character sheet. The game reports two decimals and 40.03 counts. 100 is the ceiling and nothing trains past it.
4. Fill Levels gained per in-game hour. Note the level, let one in-game hour pass — two real minutes at 1× — and note it again. Measure on the method you intend to keep using: a rate taken while sparring an Inquisitor tells you nothing about a rate taken while hauling iron ore.
5. Set Game speed to the speed you play at. F2 is normal, F3 doubles it, F4 runs at five times. Only the real-time card follows this control; the in-game figure is the same at every speed.
Four cards answer together. In-game time is the stretch in game hours, with the same figure in in-game days underneath. Real time, which carries the speed you picked in its own label, is how long you sit there, and its subtitle prices the other two speeds so you can see what switching costs — the served default reads 3 min 25 s at 5×, with 17 min 4 s at 1× and 8 min 32 s at 2× below it. Share of the whole climb to 100 puts your stretch in proportion; the default 40 to 60 is 0.81% of the climb, and the card's subtitle repeats that the last ten levels alone are 91.82% of it. From your target to 100 prices everything after your target, which is the number that decides whether to keep going: 43 d 8 h 34 min from level 60, with 2 d 22 h to reach 90.
The line above the cards never disappears, whatever you type: everything below level 90 is 8.18% of the climb, the last ten levels are the other 91.82%, and your own stretch is some share of that. A second line always sits under it, pricing the ten levels after your target — at the default it reads 8 h 23 min for 60 to 70. Push the target past 90 and it swaps to a warning that a single tick of experience can round down to nothing, and a third line appears only for Field Medic and Robotics, which level on time spent instead.
Four tables sit under the results. Where the grind actually is re-prices the same climb ten levels at a time at your measured speed, so the shares and the multiples stay put while only the times move. Fighting someone stronger is the Stronger Opponent Logic ladder. Which race trains this skill fastest re-runs your stretch for all nine races on the skill you picked. Training rates you control folds open into the rate ladders for the skill's own actions. Below all four, the collapsed The arithmetic, step by step prints the five-line substitution with your own numbers, which is what to paste into a forum reply when someone disputes the total.
Worked stretches, straight from the calculator
The served default — 40 to 60 Strength on a Greenlander
Strength, Greenlander, 3.3 levels gained in the in-game hour that started at level 40, target 60. The stretch comes out at 8 h 32 min of game time, which is 0.36 in-game days, and 3 min 25 s of real time at 5× — 17 min 4 s if you leave the game at normal speed.
That twenty-level climb is 0.81% of the whole 0-to-100 grind. The fourth card is the one that changes plans: from 60 up to 100 is 43 d 8 h 34 min at the same measured speed, and only 2 d 22 h of that gets you to 90.
Those figures are the reason this stretch is the default. A YouTube training guide measured 40.03 to 60 Strength at 8 hrs 32 mins of game time, and feeding its rate into the curve returns the same duration to the minute.
The last ten levels, priced twice
A character who measures 3.3 levels an in-game hour at level 90 climbs 90 to 100 in 23 h 20 min of game time: 46 min 40 s of real time at 1×, or 9 min 20 s at 5×. Those ten levels are 91.82% of the entire climb. Narrow it to 95 to 100 and the stretch is 4 h 5 min — 1 min 38 s at 5× — and still 84.17% of the climb, which makes those five levels 10.29× the whole 0-to-90 grind put together.
The same ten levels priced from a level-40 measurement look nothing like that. Keep the default rate of 3.3 levels an hour measured at 40 and the 90-to-100 row of Where the grind actually is reads 40 d 9 h 37 min, or ×835.55 the cost of the first ten levels. Both numbers are correct and they answer different questions: 23 h 20 min is what the climb costs a character who is already earning 3.3 levels an hour up there, and 40 d 9 h 37 min is what it costs someone still earning at their level-40 pace.
The practical reading is the second one. Nobody's rate survives the 90s intact, which is why the wiki's walking ceiling lands at 96.66 rather than 100.
Zero to 100 from scratch
Set the current level to 0, the target to 100 and leave the measured rate at 3.3 levels an in-game hour, and the answer is 123 d 8 h 36 min of game time. On your own clock that is 4 d 2 h 41 min at 1× and 19 h 44 min 15 s at 5× — nineteen hours of real life, with the game running five times faster than normal and the character never idle.
Stop at 90 instead and the same measured rate gets you there in 10 d 2 h 14 min of game time: 8 h 4 min 28 s at 1×, or 1 h 36 min 54 s at 5×. Ninety levels for an hour and a half of real time, then eighteen hours more for the last ten.
Read the 0-to-100 row on its own terms. It describes a character who measures 3.3 levels an hour at level 0, which is a far weaker trainer than one measuring 3.3 an hour at level 40 — the multiplier is 1.00 down there against 0.36 at 40. Adding up rows priced from different starting levels does not describe one character's playthrough.
Which race trains Strength fastest
Pick Strength, leave the stretch at 40 to 60, and the race table re-runs your measured rate for all nine races, sorted fastest first:
| Race | XP multiplier | 40 → 60 |
|---|---|---|
| Shek | ×1.10 | 7 h 45 min |
| Greenlander | ×1.00 | 8 h 32 min |
| Hive Soldier Drone | ×1.00 | 8 h 32 min |
| P4 Unit | ×1.00 | 8 h 32 min |
| Skeleton | ×1.00 | 8 h 32 min |
| Soldierbot | ×1.00 | 8 h 32 min |
| Scorchlander | ×0.90 | 9 h 29 min |
| Hive Prince | ×0.80 | 10 h 40 min |
| Hive Worker Drone | ×0.80 | 10 h 40 min |
The whole spread is 2 h 55 min across a twenty-level stretch, so race decides less about Strength than the method you train it with. Switch the skill picker and the table changes: the multipliers are per skill, nine of the 32 skills carry none at all, and Soldierbot ships a 0.80 on Martial Arts that the wiki says does not actually reduce the experience it gains.
The times move with the race because the multiplier divides out of your measurement first. Time a Shek at 3.3 levels an hour and the underlying base rate is lower than a Greenlander's at the same observed 3.3, so the table pays the Shek back at 1.10 and every other row is priced from that same base.
Four twenty-level stretches, same measured rate
Every row below is a character who measures 3.3 levels in the in-game hour that starts at that row's first level. Same rate, same skill, wildly different bills:
| Stretch | In-game time | Real at 1× | Real at 5× | Share of the climb |
|---|---|---|---|---|
| 10 → 30 | 7 h 29 min | 14 min 58 s | 3 min 0 s | 0.31% |
| 40 → 60 | 8 h 32 min | 17 min 4 s | 3 min 25 s | 0.81% |
| 60 → 80 | 10 h 53 min | 21 min 46 s | 4 min 21 s | 2.35% |
| 90 → 100 | 23 h 20 min | 46 min 40 s | 9 min 20 s | 91.82% |
The in-game times barely triple across the table while the share of the climb multiplies by nearly three hundred. That gap is the whole point of measuring rather than assuming: a character who can still earn 3.3 levels an hour at 90 is doing something extraordinary, and the share column is what tells you so.
The share column never moves with your inputs. 0 to 90 is 8.18% of the climb whatever rate you type, and 90 to 100 is 91.82% of it, because both come from the curve rather than from your measurement.
The closed form behind every figure on the page
- = A skill level. Kenshi tracks it with a decimal, so 40.03 is a real value and the form accepts it
- = The experience multiplier at that level. It reaches exactly zero at 101, which is why no skill trains past 100
- = What the stretch from a to b costs, in units where 1 unit is what one level costs at level 0. The whole climb 0 to 100 is 10,100 units
- = Your current level and your target level
- = The levels you gained in one in-game hour, measured starting at f
- = The experience you earn per in-game hour, recovered from that measurement by pricing the span you actually climbed
- = In-game hours the stretch takes. One in-game day is 24 of them
- = Game speed: 1, 2 or 5. One in-game hour is two real minutes at 1×, so an in-game day is 48 real minutes
The multiplier is a smooth curve rather than a table of rungs, so the cost of a stretch integrates in closed form. Cost per level goes as 1 ÷ m(L), and integrating that between two levels gives
which is where every share on the page comes from. Put a = 0 and b = 100 and the total is 101² × (1 − 1/101) = 10,100 units. Put a = 90 and the remainder is 91.82% of that total. Put a = 95 and it is 84.17%. Put a = 99 and the last single level is 50.50%, more than the ninety-nine below it added together.
The calibration step is the one worth understanding, because it decides what your answer means. You measured r levels in one in-game hour starting at level f, so what that hour bought was the span from f to f + r — and pricing that span with the same integral gives the experience you earn per hour. At the served default the span is 40 to 43.3, worth 9.56 units, and the stretch 40 to 60 is worth 81.58 units, so the answer is 81.58 ÷ 9.56 = 8.53 in-game hours, printed as 8 h 32 min. Real time is the last line: 8.53 in-game hours × 2 real minutes ÷ 5 = 3 min 25 s.
The consequence to keep in mind is that the rate is measured at your current level, not at your target. Type 3.3 levels an hour at level 0 and you have described a character earning 3.41 units an hour; type the same 3.3 at level 40 and you have described one earning 9.56, nearly three times as much, because the same three levels cost three times more up there. Two rows of this page priced from different starting levels are two different characters, and their times do not add up into one playthrough.
Race enters the same step and leaves it again. A measured rate already contains the racial multiplier of the character you timed, so the engine divides it out to get a race-neutral base rate before it prices anything. The same logic is why the form has no field for the enemy you are fighting: whatever Stronger Opponent Logic multiplier was in play during your measured hour is baked into r already, and asking for it a second time would count it twice. A mod that rescales experience behaves the same way — it multiplies the rate you measure, so the calibrated answer still holds.
Getting more out of the same hours
- Measure on the method you are going to leave running. The rate is the only input the curve cannot supply, and it is method-specific. Sparring gives one figure, hauling ore another, and a rate borrowed from a guide describes that guide's character rather than yours.
- For Strength, the ladder is worth eleven times the bottom rung. A weapon needing 20 or more Strength than you have trains at 110% of the skill's base rate, martial arts at 100% encumbrance at 100%, walking at 70%+ encumbrance while carrying a body at 50%, walking overloaded with empty hands at 25%, and a weapon whose requirement you already meet at 10%. Players in Steam's "Fastest way to get strength from 99 to 100?" thread rate the carrying trick highest for exactly this reason — it is "by far the best strength rate" and it automates.
- Fight above your weight and read the ladder before picking a target. Stronger Opponent Logic pays ×2.00 at 10 levels above you and ×6.00 at 50 or more, and being outnumbered stacks 10% per extra enemy up to ×1.667. A ×6 rate divides the hours by six and no further, though: 91.82% of a climb split six ways is still worth more than the entire stretch below level 90.
- Lower your effective skill instead of raising your opponent's. A Steam moderator puts the mechanic plainly: "by lowering your effective skill level with equipment, you can increase your bonus or reduce your penalty." Heavy armor with a skill penalty widens the gap the ladder reads, which is how players keep training after they have outgrown everything nearby.
- For Dexterity, the weapon decides everything. Dexterity experience is the cut share of a weapon's damage, cut ÷ (cut + blunt), so a katana trains at 100%, a rusted junk polearm at 79%, and a jitte at 0% — swing a jitte all day and Dexterity does not move. Lower-quality blades pay more than pristine ones.
- Keep a town loaded when you grind above 90. The wiki's 96.66 Strength ceiling was reached by a Shek walking with no town loaded, and the Steam thread about progress freezing at 92 ends with the player parking a second character near an NPC town to unstick the first. Above 90 an award can round down to nothing, and that is where the difference shows.
- Check the fourth card before you commit to a target. "From your target to 100" prices what is left after you stop. From level 60 at the default rate it reads 43 d 8 h 34 min, with only 2 d 22 h of that reaching 90 — which usually reframes the question from "how do I get to 100" to "which of my six characters gets to 90".
- Two characters at 70 beat one at 90, and the share column proves it. Everything below 90 is 8.18% of a climb, so a second character reaching 90 costs a fraction of pushing the first one to 100. Steam's "How do you get combat skills past ~70?" thread reaches the same conclusion from experience rather than from arithmetic.
- Switching to 5× does not shorten the grind, only your evening. The in-game figure is identical at every speed. F4 divides the real-time card by five, which turns a 48-minute in-game day into 9 min 36 s of your own — and turns any pathing accident into five times the damage before you notice.
Seven things players get wrong about Kenshi training
- "The bar is bugged past 90." It is doing what the formula says. Experience is scaled by (1 − level ÷ 101) squared, so a level at 95 costs 256 times a level at 5 and the last ten levels are 91.82% of the whole climb. A bar that has not visibly moved in a real hour can still be filling on schedule.
- "Training stopped working, so I need a mod." Two separate things happen above 90 and only one is a defect. The curve makes each level enormous, and on top of that a small enough award rounds down to zero — the case the wiki's 96.66 Strength ceiling documents. A mod that multiplies experience changes the rate you measure, and the curve keeps its shape underneath.
- "My race makes this faster, so the calculator should ask for it." It asks, and then it divides the answer back out. Your measured rate already contains your race's multiplier, so applying it again would double-count it. The picker marks your row in the comparison table; your own time comes from the hour you timed.
- "An in-game hour is roughly an hour." One in-game hour is two real minutes at normal speed, so a full in-game day costs 48 real minutes, 24 at 2× and about 9 min 36 s at 5×. Ten in-game days of grinding — the whole climb from 0 to 90 at the default rate — is an hour and a half of your own at 5×.
- "Training dummies will carry me to a decent skill." Players in Steam's thread "Do the dummies have a max level of skill you can only reach?" put the MK II at level 10 and the MK III at 15, and note that the dummy raises Melee Attack rather than a weapon skill. They exist to get a recruit good enough to fight starving bandits.
- "100 is a target." For most skills it is a limit the curve approaches rather than a level players reach. A Steam moderator's own long-running character sat around 96 Strength, another poster reports 99.80 refusing to move "for hours", and the wiki's walking ceiling is 96.66. Field Medic and Robotics are the two exceptions, and they pass 101 because they level on time spent rather than on actions.
- "Adding up the band times gives me my playthrough." Each band in "Where the grind actually is" is priced from the single rate you measured, and every worked stretch on this page assumes the rate was measured at that stretch's starting level. Nobody keeps a level-40 rate at level 90, so the bands are a map of where the cost sits rather than a schedule.
What this calculator will not tell you
Each exclusion below is a decision, and each one marks where a figure on this page stops being reliable.
It cannot tell you your rate. Per-action experience values are not published anywhere: they sit in the FCS, which needs the game installed, and no wiki page or guide lists them. Any tool claiming to compute Kenshi training time from a dropdown of actions is inventing the numbers underneath. Measuring one in-game hour takes two real minutes at 1× and gives an answer specific to your character, your gear and your opponent.
The rate is measured at your current level, so rows are not a journey. A character earning 3.3 levels an in-game hour at level 90 is doing something a character earning 3.3 at level 40 cannot. Every stretch here is priced from the rate you typed and the level you typed it at, and stretches with different starting levels describe different characters.
Field Medic and Robotics are outside the model. They level on time spent rather than on actions, and they are the only two skills that pass 101. The curve still describes their shape, and no measured rate makes the arithmetic here a stopwatch for them.
Rounding above 90 makes real play slower than any curve predicts. Once an award is small enough to round down to zero, the skill stops moving entirely rather than moving slowly, and the player in Steam's thread about progress freezing at 92 lost over 100 in-game days of forging to it. Treat every figure above as a floor once a skill is in the 90s.
Stronger Opponent Logic is a reference table, not an input. Applying it to a rate that already contains it would double the multiplier. The ladder is on the page so you can decide what to fight next; the answer follows from what you actually measured.
Mods and version. Everything here is the unmodded curve at Kenshi 1.0.68, with the racial matrix, the ladder and the rate figures checked against the wiki in August 2026. A mod that rescales experience multiplies your measurement and leaves the calibration valid; a mod that reshapes the curve itself does not, and no figure on this page survives it.
Kenshi stat training — frequently asked questions
Why does my Kenshi skill stop going up after 90?
Kenshi's experience curve is doing that. Every award is scaled by (1 − level ÷ 101) squared, so one level at 95 costs 256 times a level at 5, and levels 90 to 100 are 91.82% of the whole climb. A bar filling on schedule looks frozen up there.
How long is a day in Kenshi?
Forty-eight real minutes at normal speed, so one in-game hour is two real minutes. F3 doubles the clock and drops a day to 24 minutes, and F4 runs at five times, which is about 9 minutes 36 seconds per in-game day.
How long does it take to train Strength to 100 in Kenshi?
For a character measuring 3.3 levels in an in-game hour at level 0, the full 0-to-100 climb is 123 d 8 h 36 min of game time — 4 d 2 h 41 min of real time at 1×, or 19 h 44 min 15 s at 5×. Reaching 90 instead costs 10 d 2 h 14 min of game time, about an hour and a half of real time at 5×. Nobody holds one rate across that whole range, which is why the honest answer is to measure your own hour and price the stretch in front of you.
Is 100 in a skill actually reachable without mods?
For most skills, no. The multiplier hits zero at 101 and the awards round down to nothing well before that: the wiki records a hard ceiling of 96.66 Strength for a Shek training by walking with no town loaded, and players report 99.80 refusing to move for hours. Field Medic and Robotics are the exceptions and can pass 101, because they level on time spent rather than on actions.
Which race trains a skill fastest in Kenshi?
It depends on the skill, and the spread is smaller than most players assume. For Strength the fastest is Shek at ×1.10 and the slowest is Hive Prince and Hive Worker Drone at ×0.80, which across a 40-to-60 stretch is 7 h 45 min against 10 h 40 min. Nine of the 32 skills carry no racial modifier at all, so on those every race trains at the same speed.
Do I have to enter my race for the result to be right?
Enter the race of the character you measured, because the engine divides that multiplier back out to recover a race-neutral base rate. It never multiplies your answer. Picking the wrong race changes which row the comparison table marks and shifts the other rows around your figure.
Why is there no field for the enemy I am fighting?
Because your measured rate already contains it. Stronger Opponent Logic multiplied whatever experience you earned during the hour you timed, so asking for the opponent gap again would apply the multiplier twice. The ladder is printed on the page as a reference for choosing your next target.
Does fighting stronger enemies really train faster in Kenshi?
Yes, and the ladder is steep. Fighting 10 levels above you pays ×2.00, 30 above pays ×4.00 and 50 or more pays ×6.00, while an opponent 22.5 levels below you pays ×0.10. Being outnumbered adds 10% per extra enemy up to ×1.667, so the ceiling is ×10. Only eleven skills read the ladder — Melee Attack, Melee Defence, Martial Arts, Dodge, Dexterity and the weapon skills — so a crafting or Athletics grind gains nothing from a harder fight.
Does running the game at 5x speed make training faster?
It shortens your evening and leaves the grind exactly the same size. The in-game time is identical at every speed; only the real-time card changes, dividing by two at F3 and by five at F4. A stretch costing 8 h 32 min of game time is 17 min 4 s of your life at normal speed and 3 min 25 s at five times.
Should I push a character past 90 or start training a second one?
The share column answers it. Everything below level 90 is 8.18% of a full climb and the last ten levels are the other 91.82%, so bringing a second character to 90 costs a small fraction of dragging the first one to 100. Players in Steam's "How do you get combat skills past ~70?" thread reach the same verdict, arguing two characters at 70 are worth more than one at 90. Push a single character into the 90s when the fight you need them for demands it, not as a default plan.
Do training dummies work for anything past the early game?
Barely. Players in the Steam thread "Do the dummies have a max level of skill you can only reach?" put the MK II at level 10 and the MK III at 15, and point out the dummy raises Melee Attack rather than a weapon skill. They are for getting a fresh recruit ready to fight the weakest enemies, and real combat overtakes them almost immediately.
How accurate is this Kenshi training calculator?
The curve and the constants are exact and come from Kenshi 1.0.68: the (1 − level ÷ 101) squared multiplier, the 10,100-unit total for a full climb, the 8.18% and 91.82% split at level 90, the Stronger Opponent Logic ladder and the nine-race matrix, all checked against the wiki in August 2026. The one uncertain input is the rate, which is why you measure it rather than pick it from a list. Out of scope and stated on the page: Field Medic and Robotics, the rounding that freezes skills above 90, and any modded experience curve.
Is this Kenshi calculator free to use?
Yes, with nothing held back. All four cards, the four reference tables and the step-by-step arithmetic render on the first load, with no account and no limits, and it embeds into a guide of your own if you want it there.
Sources & References
- Kenshi Wiki, "Guide to Training Statistics" — the experience formula this page integrates: XP is multiplied by (1 - level/101) squared, so the multiplier reaches zero at 101 and training slows without ever stopping. Also the source of the encumbrance and weapon-requirement rate ladders, and of the wiki's own observation that the top Strength method is eleven times the bottom one.
- Kenshi Wiki, "Stronger Opponent Logic" — the eleven-step multiplier ladder for fighting someone above or below your own skill, the eleven skills it applies to, and the outnumbered bonus with its 1.667x cap at one against eight. The page's own per-enemy ladder reaches 1.7x there; this calculator uses the cap, because that is the figure consistent with the 10x ceiling the same page states.
- Kenshi Wiki, "Statistics" — the 9 x 32 racial experience matrix, read through the wiki's API. All 288 cells were diffed programmatically against the live wikitext before shipping, with zero mismatches; nine of the 32 skills carry no racial modifier at all, and the nine identical rows the table then shows are the correct answer rather than a gap.
- Kenshi Wiki, "Strength" — the per-method training rates: a weapon needing 20+ more Strength than you have trains at 110%, martial arts at full encumbrance at 100%, carrying a body at 50%, walking overloaded at 25%, and a weapon you already meet the requirement for at 10%.
- Kenshi Wiki, "Toughness" — read for the racial spread this calculator's table reports (Shek and all three Hive types train it at 1.2x) and to confirm that Toughness follows the same level curve as every other skill rather than a special case.
- Kenshi Wiki, "Athletics" — running at zero encumbrance trains at 150%, the fastest single rate in the table, and the page also documents the encumbrance penalty this calculator reproduces for Dodge.
- Kenshi Wiki, "Dexterity" — that Dexterity XP is the weapon's cut damage divided by its total cut plus blunt, which is why a katana trains it fully, a rusted junk polearm at 79%, and a jitte not at all. That last zero is a value, not a gap.
- Kenshi Wiki, "Encumbrance" — the 1% Dodge XP lost per 1% of encumbrance and the floor at 60%. A second wiki page implies no floor at all; this calculator ships the version that names a number, and the table's caption says so.
- Kenshi Wiki, "Global Game Settings" — the game version these figures are documented for (1.0.68) and the global experience multiplier. A mod that changes XP multiplies the speed the player measures, so the measured rate absorbs it and the answer still holds.
- Kenshi Wiki, "Training Dummy" — checked to rule dummies out of the curve: they cap the skill level they can train rather than changing the rate at which XP accrues, so they bound where this calculator's answer stops applying rather than altering it.
- Kenshi Wiki, "Controls" — the vanilla game-speed keybinds, which is what settles the three speeds this calculator offers: F2 normal, F3 double, F4 five times. The Product spec had drafted them from a mod description and flagged them unverified; this page is the game's own control table.