Qwen 2 57B A14B: benchmark results
Alibaba's 57B mixture-of-experts Qwen2 with 14B active per token, delivering roughly 30B-dense performance at lower compute (June 2024). Provider: Alibaba. Released 2024-06-06. Access: Open.
Aggregate Rating 1499 ± 21, rank #783 of 1656 rated models, from 97 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Portuguese LLM - ASSIN2 RTE | 94.44 | Macro F1 (%) | 96.5 |
| Open Japanese LLM - SUM | 12.17 | Score (%) | 92.5 |
| Open Japanese LLM - Xlsum JA Rouge2 | 12.18 | Score (%) | 92.5 |
| Open Japanese LLM - Xlsum JA Rouge1 | 33.43 | Score (%) | 91.8 |
| Open Japanese LLM - Xlsum JA RougeLsum | 28.16 | Score (%) | 91.8 |
| Open Japanese LLM - Xlsum JA Bert Score JA F1 | 70.97 | Score (%) | 91.6 |
| Open Portuguese LLM - OAB Exams | 57.77 | Accuracy (%) | 90.9 |
| Open Chinese LLM - GSM8K | 64.9 | Accuracy (%) | 89.3 |
| Open LLM Leaderboard - MMLU-Pro | 49.16 | Score | 87.3 |
| Open Chinese LLM - C-Eval Semantic | 87.95 | Accuracy (%) | 87 |
| Open Portuguese LLM - ENEM | 74.25 | Accuracy (%) | 86.4 |
| Open Portuguese LLM - BLUEX | 65.23 | Accuracy (%) | 86.2 |
Interactive version: theaggregate.ai/model?slug=qwen-2-57b-a14b · How It Works · Data refreshed daily, snapshot 2026-10-20.