SecCodeBench — leaderboard
Security benchmark for AI-generated and AI-repaired code, reporting secure-code repair and generation scores with and without hints.
Metric: Total Score. Source: alibaba.github.io. Status: saturation imminent. 36 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 | 68.1 |
| 2 | Qwen 3.5 Plus (Thinking) | 67.08 |
| 3 | Claude Opus 4.6 | 64.9 |
| 4 | Qwen3 Coder Next | 62.76 |
| 5 | Gemini 3 Pro | 62.42 |
| 6 | GLM-5 (Thinking) | 62.13 |
| 7 | Kimi K2.5 (Thinking) | 61.25 |
| 8 | GPT-5.4 | 59.74 |
| 9 | Gemini 3 Flash | 58.66 |
| 10 | GPT-5.2 | 58.23 |
| 11 | Claude Sonnet 4.5 | 56.83 |
| 12 | DeepSeek V3.2 (Thinking) | 55.24 |
| 13 | Kimi K2.5 (Non-reasoning) | 55.22 |
| 14 | Gemini 3.1 Pro (Preview) | 55.21 |
| 15 | DeepSeek R1 0528 | 54.06 |
Interactive version: theaggregate.ai/benchmark?slug=seccodebench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.