Qwen-72B: benchmark results
Alibaba Cloud's Qwen-72B, the base pretrained checkpoint in the Qwen series, trained on over 3 trillion tokens of Chinese, English, multilingual, code and math data, with 32K context support. Provider: Alibaba. Released 2023-11-26. Access: Open.
Aggregate Rating 1500 ± 22, rank #777 of 1635 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CFLUE | 72.8 | Accuracy | 100 |
| Open LLM Leaderboard v1 - MMLU | 77.37 | Accuracy (%) (5-shot) | 98.7 |
| Open LLM Leaderboard v1 - GSM8K | 70.43 | Accuracy (%) (5-shot) | 93 |
| Open Portuguese LLM - OAB Exams | 60.32 | Accuracy (%) | 92.6 |
| Open Portuguese LLM - ENEM | 78.24 | Accuracy (%) | 91.9 |
| Open Portuguese LLM - BLUEX | 68.43 | Accuracy (%) | 91.2 |
| InfiBench | 55.34 | Score (%) | 85.7 |
| Open Portuguese LLM - ASSIN2 RTE | 92.88 | Macro F1 (%) | 84.7 |
| Open LLM Leaderboard v1 - WinoGrande | 82.48 | Accuracy (%) (5-shot) | 84.3 |
| C-Eval | 82.8 | Average (%) | 83.8 |
| Open Portuguese LLM - FaQuAD NLI | 78.4 | Macro F1 (%) | 83.7 |
| T-Eval | 71.4 | Overall Score (%) | 80 |
Interactive version: theaggregate.ai/model?slug=qwen-72b · How It Works · Data refreshed daily, snapshot 2026-11-05.