Qwen 2 72B Instruct: benchmark results
Alibaba Qwen 2 72B instruction-tuned checkpoint. Provider: Alibaba. Released 2024-06-06. Access: Open.
Aggregate Rating 1562 ± 7, rank #576 of 1656 rated models, from 889 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM EWoK - Social Relations | 99.49 | Exact Match (%) | 100 |
| HELM EWoK - Spatial Relations | 79.59 | EM | 100 |
| HELM SeaHELM - IndoNLI | 86 | EM | 100 |
| LLM Stats (TheoremQA) | 44.4 | Score (%) | 100 |
| Open Chinese LLM - C-Eval Semantic | 94.12 | Accuracy (%) | 100 |
| Open Chinese LLM - GSM8K | 78.7 | Accuracy (%) | 100 |
| Open Chinese LLM Leaderboard | 75.67 | Official Nonzero-Task Mean (%) | 100 |
| OpenCompass Knowledge - Engineering - Chinese (CompassBench 2407) | 85.6 | Score (%) | 100 |
| OpenCompass Knowledge - Science - English (CompassBench 2407) | 92 | Score (%) | 100 |
| SeaEval - Cultural Reasoning - US-Eval (Zero-Shot) | 87.85 | Accuracy (%) | 100 |
| SeaEval - Fundamental NLP Tasks - MRPC (Zero-Shot) | 80.64 | Accuracy (%) | 100 |
| SeaEval - Fundamental NLP Tasks - OCNLI (Zero-Shot) | 78.2 | Accuracy (%) | 100 |
Interactive version: theaggregate.ai/model?slug=qwen-2-72b-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.