XOR Vulnerability Repair - Patch (Claude Code): leaderboard
XOR's automated vulnerability repair benchmark, defender role: each system patches memory-safety vulnerabilities in low-level software that have public fixes, and a withheld grader replays the trigger and rebuilds. This board holds the Claude Code harness runs from April 2026 (published September 2026); score is the share of vulnerabilities whose patch passed, 0-100, higher is better. XOR does not record the reasoning effort of these runs.
Metric: Pass Rate (%). Source: www.xor.tech. Saturation forecast: Around June 2027. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 56.9 | #51 |
| 2 | Claude Opus 4.6 | 46.2 | #67 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=xor-vulnerability-repair-patch-claude-code · How It Works · Data refreshed daily, snapshot 2026-11-05.