Global Leaderboard
Per-Language
Model Status
Best model per language — ranked by the average of WER and CER (lower is better). Individual WER/CER scores shown for transparency.
Open models are the core of this benchmark: their weights can be downloaded and the scores re-run by anyone. The closed models and LLM tracks are shown for comparison.
| # | Language | Model | Type | Track | Params |
|---|
✅ Passed
| # | Model | Type | Track | Params | License |
|---|
❌ Failed
| # | Model | Type | Track | Params | License | Reason |
|---|
| Language | Code | Family | Model | Type | Track | Params | License | Avg (WER+CER) (%) | WER (%) | CER (%) | Status | Fail Reason |
|---|