aiXamine - Code Security: leaderboard
aiXamine Code Security service score: avoiding vulnerable code in instruction and autocomplete settings on CyberSecEval 3 and SecCodePLT (aiXamine, arXiv 2608.20554v1, Appendix Table 7). Normalized 0-100 score aggregated from the service's tests; higher is better. 42 models evaluated under identical black-box conditions via OpenRouter or containerized vLLM.
Metric: Service score (0-100; normalized pass rate). Source: arxiv.org. Saturation forecast: Around June 2027. 42 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-OSS-120B | 80.1 | #346 |
| 2 | GPT-5 Mini | 79.7 | #193 |
| 3 | GPT-OSS-20B | 78.2 | #523 |
| 4 | DeepSeek R1 0528 | 77.1 | #232 |
| 5 | Gemma 4 26B A4B | 76.4 | #360 |
| 6 | Gemma 4 31B | 75.3 | #212 |
| 7 | Qwen 3.5 27B | 75 | #192 |
| 8 | Gemini 3 Flash (Preview) | 74.3 | #90 |
| 9 | Qwen 3.5 122B A10B | 74.1 | #184 |
| 10 | Qwen 3.5 9B | 73.2 | #391 |
| 11 | GPT-4o | 73.1 | #357 |
| 12 | Gemini 2.5 Flash | 72.2 | #255 |
| 13 | Claude 3.5 Haiku | 72.2 | #571 |
| 14 | Qwen 3.5 35B A3B | 72.1 | #268 |
| 15 | Grok 4.1 Fast | 72 | #227 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aixamine-code-security · How It Works · Data refreshed daily, snapshot 2026-11-05.