Qwen 2.5 32B: benchmark results
Alibaba's 32B dense base model from the Qwen2.5 series, pretrained on up to 18T tokens under Apache 2.0 (September 2024). Provider: Alibaba. Released 2024-09-19. Access: Open.
Aggregate Rating 1570 ± 9, rank #546 of 1656 rated models, from 342 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open PL LLM - PPC (generative, 5-shot) | 83.6 | Accuracy (%) | 100 |
| Open PL LLM - PPC (multiple choice, 5-shot) | 83.9 | Accuracy (%) | 100 |
| Open LLM Leaderboard - GPQA | 41.19 | Score | 99.6 |
| Open Portuguese LLM - FaQuAD NLI | 86.86 | Macro F1 (%) | 99.6 |
| Open LLM Leaderboard - MMLU-Pro | 58.05 | Score | 99.4 |
| Open PL LLM - PSC (multiple choice, 5-shot) | 94.59 | Binary F1 (%) | 99.4 |
| Open PL LLM - DYK (generative, 5-shot) | 74.15 | Binary F1 (%) | 99.1 |
| Open PL LLM - DYK (multiple choice, 5-shot) | 73.03 | Binary F1 (%) | 99.1 |
| Open PL LLM - KLEJ NER (generative, 5-shot) | 61.9 | Accuracy (%) | 99.1 |
| Open PL LLM - Multiple Choice | 64.7 | Average Multiple-Choice Score (%) | 98.3 |
| Open PL LLM - Belebele (generative, 5-shot) | 92.56 | Accuracy (%) | 98.1 |
| Swallow - Pre-trained English - MMLU | 83.2 | Accuracy (%) | 98 |
Interactive version: theaggregate.ai/model?slug=qwen-2-5-32b · How It Works · Data refreshed daily, snapshot 2026-10-20.