ConlangBench
Explore score history ↗ConlangBench evaluates large language models on translation and vocabulary learning across 21 constructed languages, with a corpus of over 21 million conlang-English parallel sentence pairs and 321K vocabulary entries. The benchmark includes bidirectional translation tasks and learning-curve analysis for models trained on eight conlangs with sufficient parallel data.
- Released
- 2026-08-04
- Institutions
- Yonsei University · Seoul National University
- Evaluates
- Factuality, General AI, Language & Knowledge, cs.CL
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
ConlangBench paper
No reported scores are on record for this benchmark yet.