Qwen-14B: benchmark results
Alibaba Cloud's Qwen-14B, the base pretrained checkpoint in the Qwen series, trained on over 3 trillion tokens of Chinese, English, multilingual, code and math data. Provider: Alibaba. Released 2023-09-25. Access: Open.
Aggregate Rating 1415 ± 19, rank #1171 of 1656 rated models, from 36 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - MMLU | 67.7 | Accuracy (%) (5-shot) | 89.8 |
| ARC Challenge (AI2) | 84.4 | Accuracy (%) | 81.6 |
| CFLUE | 53.82 | Accuracy | 81.2 |
| BoolQ | 86.2 | Accuracy (%) | 76.6 |
| T-Eval | 66.3 | Overall Score (%) | 75 |
| Open LLM Leaderboard v1 - GSM8K | 58.98 | Accuracy (%) (5-shot) | 70.7 |
| InfiBench | 43.69 | Score (%) | 69.5 |
| GSM8K | 61.3 | Accuracy (%) | 65.6 |
| C-Eval | 72.1 | Average (%) | 63.4 |
| Open LLM Leaderboard v1 - HellaSwag | 83.99 | Normalized accuracy (%) (10-shot) | 63 |
| EvoEval Difficult | 23 | Pass@1 (%) | 56 |
| Big-Bench Hard | 53.4 | Average (%) | 54.2 |
Interactive version: theaggregate.ai/model?slug=qwen-14b · How It Works · Data refreshed daily, snapshot 2026-10-20.