My Pals

How we check the data

For Steam build 25247047 our engine gives the same child as all 4 breeding tables we test against, on every row, and fits 6 of 7 in-game hatch reports.

Build 25247047 (Palworld v1.0.5)

Game version
v1.0.5
Steam build
25247047
Patch date
Game files extracted
Site data generated
  • 300Pal entries
  • 288obtainable by breeding
  • 81special combos
  • 23self-only Pals
  • 11Terraria Pals
  • 44,850parent pairs checked

Data changes per build are listed in the changelog, game patches in updates.

How we check the data

Every time we ingest a build, the pipeline runs our breeding engine against each table below. The build stops if a single row disagrees.

Their breeding results are used only as a test and never shipped. Pal Calc (MIT licence) also supplies names, learnsets, passives and icons; see sources.

Different people built these tables from their own copies of the game data, and they probably all follow the same community reading of the game code.

A full match shows that our data and code agree with theirs, which does not prove the game behaves that way.

The direct evidence is in-game hatches: 6 of the 7 reports we have fit the rule. The one that does not is explained under known disagreements.

Harness results for build 25247047
Checked againstWhat is comparedResult
Pal Calc Its full breeding table, one row per parent pair and gender case. 44,851 of 44,851 rows match
paldb.cc Every parent pair its breeding pages list. 44,253 of 44,253 pairs match
palworld.gg Its calculator table, one row per pair and gender combination. 88,506 of 88,506 rows match
pindrop.gg The pairs listed on its per-Pal breeding pages. 2,878 of 2,878 pairs match
Golden cases 7 in-game hatch reports and 5 checks of special cases in the data, such as the gendered and removed recipes. Listed below. 12 of 12 match the data (1 contradicted by an in-game report, pending retest)
Mutation targets Which species a mutation can produce, against the species paldb.cc flags as mutation results. 143 species (paldb flags 143; 0 differ)
Changed since 0.7.3 Pairs of Pals that existed before 1.0 whose child is different now. See 1.0 breeding changes. 22,929 of 23,436 pairs (97.8%)

In-game reports and recipe checks

Golden cases, re-run with the engine on this page's build
ParentsEngine givesKindEvidence
Kitsun Noct + Felbat Bulldosu In-game report Pal Calc issue #190, in-game: on a breeding power tie, the Pal with the higher breeding priority wins
Fuddler + Suzaku Souffline In-game report Pal Calc issue #190, in-game
Jormuntide Ignis + Braloha Wumpo Botan In-game report Pal Calc issue #190, in-game
Elgrove Cryst + Mycora Wumpo Botan In-game report Pal Calc issue #196, in-game
Anubis + Beakon Relaxaurus In-game report Reddit thread, in-game: Pals from special recipes are left out of the breeding power match
Penking + Bushi Sibelyx In-game report Reddit thread, in-game (implied by the same report)
Katress (male) + Wixen (female) Wixen Noct Data check Special recipe that depends on gender
Katress (female) + Wixen (male) Katress Ignis Data check Special recipe that depends on gender
Wixen (female) + Katress (male) Wixen Noct Data check Special recipe that depends on gender, parents swapped (order does not matter)
Grizzbolt + Relaxaurus Reptyro Data check 1.0 removed the Orserk recipe. The target breeding power (1055) sits halfway between Ragnahawk and Reptyro, and Reptyro wins on breeding priority
Jetragon + Lamball Bushi Data check Pals that cannot be bred still work as parents
Warsect + Digtoise Warsect Terra (one report says Cryolinx) In-game report Special recipe in the game data

Sources and licences

Data sources for this build
SourceUsed forVersionLicence
palworld-atlas-data primary: roster, breeding power, special recipes, spawns, items (Steam build 25247047) feb0ca6 MIT
Pal Calc enrichment (localized names, learnsets, passives, icons) and breeding oracle 8566b9a MIT
CreativeTechGuy/PalworldDBIndex raw DataTables breeding exclusions, priorities, Terraria rows, partner skills, drops, fishing, game settings; 0.7.3 tables for legacy data 862463c none stated (game facts only, files not redistributed)
Dhampyru/Palworld-Extracted cake recipes, world presets (hatch time), item and map-object names snapshot of 2026-07-22 none stated (game facts only, files not redistributed)
Steam news (Palworld app 1623730) patch date v1.0.5 n/a (dates only)
catrenelle/PalDex Map images we cut into tiles, and map points: fast travel, watchtowers, towers, bounties, oil rigs, dungeons, quests and collectibles snapshot of 2026-08-03 MIT
Nifrendil/pal-atlas Map points: Sealed Realms, plus a cross-check of fast travel and tower positions snapshot of 2026-08-02 MIT
LukeHollandDev/palworld-live-map Map points: extra dungeon entrances and effigy names, plus a cross-check of fast travel, watchtowers, bounties and oil rigs game 1.0.3 MIT

Copyright lines and the full licence notices are on the credits page.

Where a source states no licence we use only game facts from it (numbers and names) and do not redistribute its files. Pal names, icons and game text belong to Pocketpair.

From game files to pages

  1. A new Steam build of the game is extracted into data tables. We pin the exact revision of every source above.
  2. The pipeline reads the roster, breeding power, special recipes and spawns, and fills gaps (tie-break priorities, partner skills, drops, game settings) from the raw data tables.
  3. It sorts every Pal into the breeding pool or out of it, with the reason. See the breeding pool.
  4. It writes one compact data file for the breeding engine and one JSON file per topic for the pages.
  5. It runs the checks above. Any mismatch, or a change in the expected counts, stops the build so a person looks at it.
  6. The site is rebuilt from those files. The breeding calculator, the breeding chart and these guides all run the same engine code on the same data.

The breeding rule in plain words

Datamined code For two parents, the game picks the child like this:

  1. If the two parents form one of the 81 special combos, the egg is that combo's child. Only Katress with Wixen depends on which parent is male.
  2. If both parents are the same Pal, the egg is that Pal.
  3. Otherwise it averages the two parents' breeding power, rounding half up, and picks the closest Pal in the pool. The pool of 184 Pals leaves out special combo children, self-only Pals and a few others.
  4. If two Pals are equally close, the one with the higher tie-break priority wins. This decides 31% of the pairs that breeding power decides.

Game data Every breeding power in Palworld v1.0.5 is a multiple of 10, so the sum of two is always even and the rounding never changes a result.

The long version with an interactive example: how breeding works.

Confidence labels

Numbers outside the main breeding table carry one of these labels. Each label links here.

Game data
Read straight from a game data table, such as breeding power, cake recipes or world settings.
Datamined code
Taken from the game code rather than a table, for example the mutation rate or the passive inheritance steps. We trust it, but a patch can change code without changing a table.
Community-tested
Measured by players in game and not stored in any data table.
Unverified
A single report or a community figure we could not confirm. Treat it as a hint.

Known disagreements

Warsect + Digtoise

Unverified The data has a special combo for this pair, so the engine gives Warsect Terra. One in-game report says the eggs were Cryolinx.

Cryolinx has breeding power 1040, while these parents average to 1415, so no form of the breeding power rule gives it. Our best guess is a different pair in the farm.

We keep the data's answer until someone retests it.

Mutation reports

Datamined code Mutation odds follow the datamined formula; 6 of 8 player mutation reports fit it, 2 contradict it (Jetragon + Ophydia, Anubis + Nyafia). For the two that disagree, the formula gives:

Mutation outcomes the formula gives, against player reports
ParentsPlayers reportedFormula gives (share of mutated eggs)
Jetragon + Ophydia Dualith Noct and Eidrolon Ignis Eidrolon Ignis 100%
Anubis + Nyafia only Aegidron Silvegis 31%, Beakon Cryst 21%, Jormuntide 21%, Celesdir 21%, Reptyro Cryst 4.2%, Univolt Cryst 2.1%

Both reports are single anecdotes. We show mutation odds with a "Datamined code" label until an in-game test settles them.

Gumoss (Flower)

Gumoss (Flower) has the same breeding power as Gumoss and always loses the tie, so the breeding power rule never picks it. In practice you breed it from two Gumoss (Flower).

It is not counted among the 23 self-only Pals because it is still in the breeding pool. It just never wins there.

After a patch

Our target is to have data-only changes (breeding power, combos, spawns) live within one hour of a new build being detected.

New Pals go live within 24 hours, with placeholder art until the icons are extracted.

Every page shows the build it was checked against. Past patches: updates.

Report an error

If an egg hatched something other than what we show, email report@palworldbreeding.gg with the two parents, their genders, the cake you used and what hatched. A screenshot helps.

Every calculator result also has a report link that starts the email for you.

Report wrong result