Qwen2.5-Coder-0.5B-Instruct: benchmark results
Provider: Alibaba.
Aggregate Rating 1277 ± 41, rank #9003 of 10916 rated models, from 8 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LintEval - Pass@1 | 48.8 | Pass@1 (%; share of the 498 LintEval-Lite C++/Python/Java pr | 38.5 |
| Delulu - Edit Similarity | 62.5 | Edit similarity (0-100): character-level normalized Levensht | 25 |
| Delulu | 0.44 | pass@1 | 20 |
| Delulu - Exact Match | 22.8 | Exact match (%) of the completion with the gold completion, | 20 |
| Wolfram LLM Benchmarking Project | 9.1 | Correct Functionality (%) | 4.4 |
| Aider Code Editing Leaderboard | 14.3 | Exercises completed correctly after one retry, pass_rate_2 ( | 0 |
| Delulu - CodeBLEU | 36 | CodeBLEU (0-100): n-gram, syntax-tree and dataflow agreement | 0 |
| LintEval - Reduction@P | -1.2 | Reduction@P (%; corpus-level net reduction of the original l | 0 |
Interactive version: theaggregate.ai/model?slug=qwen2-5-coder-0-5b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.