Sonar LLM Leaderboard - Java - Code Smell Density: leaderboard
SonarQube code smells (maintainability issues) per 1,000 lines of generated Java code on Sonar's LLM Leaderboard, which has each model solve a fixed set of Java programming assignments and analyses the generated code with SonarQube; lower is better.
Metric: Code smells per 1,000 lines of code (lower is better). Source: www.sonarsource.com. Saturation forecast: Not forecast. 74 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 (High) | 13.79 | #87 (Claude Opus 4.5) |
| 2 | GLM-5 | 14.17 | #149 |
| 3 | Claude Sonnet 5 (Thinking) | 14.28 | #65 (Claude Sonnet 5) |
| 4 | Kimi K2.5 (Thinking) | 14.5 | #152 (Kimi K2.5) |
| 5 | GLM-4.7 | 15 | #202 |
| 6 | Mistral Large 3 | 15.02 | #410 |
| 7 | MiniMax-M2.5 | 15.08 | #316 |
| 8 | Kimi K2.6 | 15.26 | #113 |
| 9 | MiniMax-M2.7 | 15.33 | #270 |
| 10 | Gemini 4 Argon (High) | 15.53 | #9 (Gemini 4 Argon) |
| 11 | MiniMax-M3 | 15.97 | #151 |
| 12 | Claude Opus 4.6 (High) | 16.34 | #67 (Claude Opus 4.6) |
| 13 | Claude Opus 4.8 (High) | 16.71 | #39 (Claude Opus 4.8) |
| 14 | Claude Sonnet 4.5 (High) | 16.77 | #148 (Claude Sonnet 4.5) |
| 15 | DeepSeek R1 | 16.9 | #262 |
No result here: #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=sonar-llm-leaderboard-java-code-smell-density · How It Works · Data refreshed daily, snapshot 2026-11-05.