Qwen 2.5 14B Instruct: benchmark results
Alibaba Qwen 2.5 14B instruction-tuned checkpoint. Provider: Alibaba. Released 2024-09-19. Access: Open.
Aggregate Rating 1539 ± 5, rank #647 of 1656 rated models, from 802 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HiMed (English) | 61.1 | Accuracy (%; mean of the four English sets, zero-shot multip | 100 |
| Open PL LLM - EQ-Bench | 65.19 | EQ-Bench Score | 100 |
| STEMVerse - Mathematics - Bloom Level: Create | 37.5 | Few-shot accuracy (%) | 100 |
| Open PL LLM - PolEmo2-IN (generative, 5-shot) | 89.06 | Accuracy (%) | 99.4 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 75.97 | Exact match, flexible extract (%) | 99.2 |
| Open LLM Leaderboard - MATH Level 5 | 55.29 | Score | 99.1 |
| Open PL LLM - PolEmo2-OUT (multiple choice, 5-shot) | 82.39 | Accuracy (%) | 99.1 |
| Open Ko-LLM Leaderboard | 53.23 | Average (%) | 99 |
| Open Korean LLM Leaderboard | 53.23 | Average Score (%) | 99 |
| Open Korean LLM - Ko-IFEval | 66.99 | Strict accuracy, prompt/instruction mean (%) | 98.7 |
| Open LLM Leaderboard - IFEval | 81.58 | Score | 98.7 |
| Open Korean LLM - KorNAT-CKA | 39.56 | Normalized accuracy (%) | 98.6 |
Interactive version: theaggregate.ai/model?slug=qwen-2-5-14b-instruct · How It Works · Data refreshed daily, snapshot 2026-10-20.