aya-23-35B: benchmark results
Cohere Aya 23 35B multilingual instruction-tuned model. Provider: Cohere. Released 2024-05-23. Access: Open.
Unified ELO 1469 ± 1, rank #843 of 1392 rated models, from 333 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Japanese LLM - Xlsum JA Bert Score JA F1 | 72.12 | Score (%) | 97.1 |
| Open Japanese LLM - SUM | 14.43 | Score (%) | 96.9 |
| Open Japanese LLM - Xlsum JA Rouge2 | 14.41 | Score (%) | 96.9 |
| Open Japanese LLM - Xlsum JA Bleu JA | 4.15 | Score (%) | 96.6 |
| Open Arabic LLM - Alghafa Multiple Choice Rating Sentiment Task | 57.38 | Accuracy (%) | 96.3 |
| EuroEval English NLU - SQuAD | 87.4 | Reading comprehension Score (%) | 94.5 |
| EuroEval French NLU - FQuAD | 74.54 | Reading comprehension Score (%) | 93.8 |
| Open Arabic LLM - Aratrust Offensive | 94.2 | Accuracy (%) | 93.5 |
| Open Japanese LLM - Xlsum JA RougeLsum | 29.01 | Score (%) | 93.5 |
| EuroEval Portuguese NLU - MultiWikiQA PT | 77.18 | Reading comprehension Score (%) | 93.3 |
| EuroEval Spanish NLU - MLQA ES | 67.9 | Reading comprehension Score (%) | 92.4 |
| Open Japanese LLM - Xlsum JA Rouge1 | 34.08 | Score (%) | 92.1 |
Interactive version: theaggregate.ai/model?slug=aya-23-35b · How It Works · Data refreshed daily, snapshot 2026-09-05.