Grok 4.7 Leads New AI Cyber Defense Benchmark

SpaceXAI's Grok 4.7 model has tied for the top spot on the new Artificial Analysis Cyber Index with a score of 56, though it carries a premium cost of $11.67 per task.

October 4, 2026, 07:46 PM·2 views

Photo: Crypto Briefing

The Artificial Analysis Cyber Index, launched around September 28, 2026, evaluates AI agents on enterprise cybersecurity tasks, including vulnerability detection and patching. The index is supported by the Artificial Analysis Cyber Index Alliance, which includes IBM and NVIDIA.

SpaceXAI's Grok 4.7 model achieved a composite score of 56, tying for first place with MiMo-V2.6-Pro. The model demonstrated a 68% pass@1 rate on the CWE-Bench-AA test and a 74% success rate on the CyberGym-E2E-AA patching task.

Despite its performance, Grok 4.7 is noted for its cost of $11.67 per task, which is higher than that of competing models. The model follows the release of Grok 4.6, which was benchmarked earlier in September 2026.