xAI Grok 4.6 Model Card - SecureCodeReview: leaderboard
xAI's internal SecureCodeReview is an agentic software-engineering eval scoring whether an agent fixes the security issues present in a realistic codebase and whether its patch avoids merging new ones. xAI Grok 4.6 model card (revised 2026-08-17), section 7.3 chart; reward in percent, higher is better; Grok runs in the Grok Build harness.
Metric: Reward (%). Source: media.x.ai. Saturation forecast: Around September 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (xHigh) | 64.1 | #33 (GPT-5.5) |
| 2 | Grok 4.6 (High) | 58.7 | #29 (Grok 4.6) |
| 3 | GPT-5.6 Sol (Max) | 57.5 | #22 (GPT-5.6 Sol) |
| 4 | Grok 4.5 (High) | 49.4 | #54 (Grok 4.5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=xai-grok-4-6-model-card-securecodereview · How It Works · Data refreshed daily, snapshot 2026-11-05.