Qwen 2.5 0.5B Instruct: benchmark results
Alibaba Qwen 2.5 0.5B instruction-tuned checkpoint. Provider: Alibaba. Released 2024-09-19. Access: Open.
Aggregate Rating 1261 ± 11, rank #1529 of 1656 rated models, from 389 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MERA Code - CodeCorrectness | 82.73 | Accuracy (%) | 83.1 |
| SLM Trustworthiness - Fairness | 48.99 | TrustLLM fairness accuracy (%; sub-task mean) | 80 |
| Open Korean LLM - KorNAT-Helpful | 46.98 | Normalized accuracy (%) | 51.3 |
| Open Arabic LLM - Arabic MMLU HT Machine Learning | 35.71 | Accuracy (%) | 50.6 |
| MERA - ruHumanEval | 10.79 | pass@1 (%) | 48.8 |
| Open LLM Leaderboard - MATH Level 5 | 10.35 | Score | 48.5 |
| Open Korean LLM - Ko-EQ-Bench | 1.65 | EQ-Bench score | 47.2 |
| MERA - ruCodeEval | 7.56 | pass@1 (%) | 45.7 |
| Open Portuguese LLM - FaQuAD NLI | 59.88 | Macro F1 (%) | 44 |
| Open Arabic LLM - Arabic MMLU HT Formal Logic | 34.92 | Accuracy (%) | 43.5 |
| SLM Trustworthiness - Robustness | 62.42 | TrustLLM robustness accuracy (%; sub-task mean) | 40 |
| Open Korean LLM - KorNAT-Harmless | 62.15 | Normalized accuracy (%) | 39.2 |
Interactive version: theaggregate.ai/model?slug=qwen-2-5-0-5b-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.