MMLU, translated language-robustness eval, Yoruba — Distinct lang subset; lowest scores in family (0.6-0.8), consistent w/ low-resource dip.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | o1 | OpenAI | 0.8 | 1 | 2026-06-13 |
| 2 | o4-mini | OpenAI | 0.7 | 1 | 2026-06-13 |
| 3 | o3-mini | OpenAI | 0.6 | 1 | 2026-06-13 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.