Model zoo

Models you can scry with

Ids resolve through the models.yaml index. Teachers are the strongest trained models; students are the compact distillations that actually ship in the crystals.

Zoo

id / family	reveals	measured quality
`khm-latn-1.0`	Khmer text → Latin reading	59.66% exact match (forward model)
`tha-g2p-base-1.0` student	Thai → IPA	teacher at 1.73% PER (public baseline 6.37%); student gated at +4.76pp
`tha-g2p-small` student	Thai → IPA (compact)	same harness; see index
`heb-diac-1.0` student	Hebrew consonants → nikud	teacher at 16.58% DER (DictaBERT 35.6%); student parity cer_delta 0.0pp on 600 samples
`urd-diac`, `urd-g2p`	Urdu → vocalized / phonemes	G2P 14.77% CER — first learned Urdu baseline we know of
in preparation	Arabic diacritization, Persian G2P	teachers verified (Arabic 2.68% DER on SadeedDiac-25; Persian 77.34% SentenceBench HA)

Where the numbers come from

Every metric above was produced by the benchmark’s own evaluator on its full test set — Misraj’s evaluator for SadeedDiac-25, SentenceBench for Persian homographs, the fixed 1,219-sentence Kaikki set for Thai — under published, zero-skip protocols. Methodology and run IDs live with each project’s results file; nothing here is self-graded on a private split.

Conformance methodology

Before any model enters the index, an export gate writes its parity record into the artifact: the student is diffed against its teacher on held-out samples, and a model whose behavior differs from its published identity does not ship. At the runtime level, the three crystals are diffed against each other on shared golden sets (generated by the Python crystal) in CI — the release gate for gem 1.0 and every subsequent crystal release.