Granite Code 20B Instruct: benchmark results
Provider: IBM. Released 2024-05-06. Access: Open.
Aggregate Rating 1374 ± 33, rank #7047 of 10916 rated models, from 107 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Italian NLU - MultiNERD IT | 70.42 | Micro F1 excluding MISC (%) | 72.4 |
| EuroEval Portuguese NLU - HAREM | 43.68 | Micro F1 excluding MISC (%) | 58.3 |
| EuroEval Spanish NLU - CoNLL ES | 61.98 | Micro F1 excluding MISC (%) | 55.4 |
| EuroEval Icelandic NLU - MIM-GOLD NER | 48.91 | Micro F1 excluding MISC (%) | 55 |
| EuroEval German NLU - Germanquad | 52.32 | Character F1 (%) | 51.9 |
| EuroEval Faroese NLU - FONE | 61.32 | Micro F1 excluding MISC (%) | 51.2 |
| EuroEval Finnish NLU - Tydiqa FI | 61.33 | Character F1 (%) | 50.4 |
| EuroEval Norwegian NLU - NorNE NB | 61.78 | Micro F1 excluding MISC (%) | 48.9 |
| EuroEval Danish NLU - Dansk | 48.74 | Micro F1 excluding MISC (%) | 46.5 |
| EuroEval Faroese NLU - ScaLA FO | 2.5 | Matthews correlation coefficient (%) | 46.2 |
| EuroEval English NLU - CoNLL EN | 67.17 | Micro F1 excluding MISC (%) | 46.1 |
| EuroEval Norwegian NLU - NorNE NN | 58.69 | Micro F1 excluding MISC (%) | 45.9 |
Interactive version: theaggregate.ai/model?slug=granite-code-20b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.