SecRespond - Vulnerability Risk Detection: leaderboard
Metric: Detection CAP-score (%) on the Vulnerability Risk (VUL) capability dimension: the detection (discovery, evidence and root-cause attribution, 0-3 per checkpoint) score achieved over all checkpoints mapped to the dimension across the 10 SecRespond cyber ranges (averaged over three LLM judges); agents in the OpenCode harness with reasoning enabled; higher is better. Source: arxiv.org. Saturation forecast: Around June 2028. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 (Thinking) | 79.2 |
| 2 | Claude Sonnet 4.6 (Thinking) | 79.2 |
| 3 | GLM-5.1 | 78.5 |
| 4 | Claude Opus 4.7 (Thinking) | 76.2 |
| 5 | Claude Opus 4.5 (Thinking) | 72.6 |
| 6 | DeepSeek V3.2 (Thinking) | 69.6 |
| 7 | GLM-5 | 68.9 |
| 8 | Claude Sonnet 4.5 (Thinking) | 68.1 |
| 9 | Kimi K2.5 (Thinking) | 65.9 |
| 10 | MiniMax-M2.7 | 63.7 |
| 11 | Gemini 3 Flash | 62.2 |
| 12 | Qwen 3.7 Max (Thinking) | 62.2 |
| 13 | GPT-5.5 | 61.5 |
| 14 | GPT-5.4 Pro (xHigh) | 61.5 |
| 15 | Kimi K2.6 (Thinking) | 55.5 |
Interactive version: theaggregate.ai/benchmark?slug=secrespond-vulnerability-risk-detection · How It Works · Data refreshed daily, snapshot 2026-09-29.