Qwen 2.5 Coder 14B: benchmark results
Alibaba's Apache-2.0 14B code base model from the Qwen2.5-Coder series (November 2024), trained on 5.5T tokens of code-heavy data. Provider: Alibaba. Released 2024-11-12. Access: Open.
Aggregate Rating 1412 ± 17, rank #1184 of 1656 rated models, from 35 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| GSM8K | 88.7 | Accuracy (%) | 91.9 |
| Open Portuguese LLM - BLUEX | 67.87 | Accuracy (%) | 90.5 |
| Open Portuguese LLM - ENEM | 75.44 | Accuracy (%) | 89.3 |
| Open Portuguese LLM - ASSIN2 RTE | 93.38 | Macro F1 (%) | 89.1 |
| Open Portuguese LLM - FaQuAD NLI | 79.84 | Macro F1 (%) | 88.5 |
| Open LLM Leaderboard - MMLU-Pro | 45.21 | Score | 84.1 |
| Open Korean LLM - KorNAT-Helpful | 49.98 | Normalized accuracy (%) | 83.3 |
| Open Portuguese LLM - OAB Exams | 52.03 | Accuracy (%) | 81.3 |
| Open LLM Leaderboard - BBH | 58.65 | Score | 79.8 |
| Open LLM Leaderboard - MATH Level 5 | 22.51 | Score | 75.7 |
| MMLU | 75.2 | Accuracy (%) | 69.4 |
| C3-Bench (Controllable Code Completion) - Implementation Control - Pass@1 | 56.1 | Pass@1 (%; unit tests on the 1,286 instructed ICC tasks) | 67.5 |
Interactive version: theaggregate.ai/model?slug=qwen-2-5-coder-14b · How It Works · Data refreshed daily, snapshot 2026-10-20.