How Pal Breed Chart computes breeding routes

What "shortest" means

For species-only planning, every pal you start with costs 0, and a bred child costs 1 + cost(parent A) + cost(parent B). The solver relaxes every breeding combination repeatedly until no cost can improve — an exact fixpoint search over the full breeding table. It is not a greedy or beam-search approximation, and it has no depth limit.

The minimized quantity is the total number of breeding operations in the plan's fully expanded dependency tree. It is not the maximum generation depth, the expected number of eggs, or elapsed time. When two routes have equal cost, the first minimum-cost route found while scanning the breeding table is kept; there is no further tie-break. The displayed recipe deduplicates repeated intermediate species, so the number of lines you see can be smaller than the optimized step count.

Unreachable and catch-only pals

If a target is already in your starting set, it costs 0. Otherwise, if no breeding combination can be reached from your starting pals, the planner reports that plainly instead of inventing a route. Some pals cannot be produced by any breeding pair at all — those have to be caught in the wild, and their pages and lookups say so. The planner has no "catch a wild pal" step: pals you can catch belong in your starting set.

Passive-skill planning

Passive plans are ranked by summed expected breeding effort. Each breeding step's effort is 300 seconds ÷ (inheritance probability × species/gender probability), and a plan's total is its step plus both parent sub-plans. Inheritance probabilities are computed with closed-form combinatorics — no Monte Carlo sampling. The estimate covers breeding cycles only: incubation, hatching, catching, and IVs are outside the model.

Desired passives and total passive counts are tracked exactly within the model, and the passive pool's dilution is part of the math — cleaner parents can out-rank shorter-looking routes. One conservative simplification: passives you did not ask for are assumed distinct between the two parents, because only their counts are tracked. The search caps requests at four desired passives and models at most 240 states per species; within those bounds it has no beam or depth limit. Ties are broken by fewer breeding steps, then surgery cost, operation count, and final passive count.

Surgery Table finishing

Optionally, plans may finish the final target on the Surgery Table. Costs come from the gold prices extracted from the game data, empty passive slots are filled before any untracked passive is replaced, and any plan that would exceed your gold budget is rejected. Surgery is never applied to intermediate parents — only to the final pal.

Where the data comes from

The embedded dataset (v26: 299 pals, 1,905 passive skills, 44,851 breeding combinations) is generated directly from an installed copy of Palworld 1.0 by our own extraction pipeline. It was not copied from another calculator. The static per-pal pages on this site are generated by the same Rust/WebAssembly engine that powers the interactive planner, so the two can never disagree.

See the data page for the current dataset version and update policy.

Your save stays on your device

Save files you select are decompressed and parsed locally in your browser by the bundled Rust/WebAssembly engine. Pal Breed Chart does not upload the save bytes or the parsed pal collection. Details in the privacy policy.