Anthropic Opus 4.6 found three times more attack flags than Google Gemini 3 Flash in the benchmark.
| Publisher | Simbian |
| Report | The Cyber Defense Benchmark: Why Every Frontier LLM Failed |
| Published | 28 April 2026 |
| Topics | Anthropic Claude Opus 4.6, Threat Detection, AI Models, Google Gemini 3 Flash |
Published by Simbian in The Cyber Defense Benchmark: Why Every Frontier LLM Failed, 28 April 2026. The figure is taken from the report as published; the full methodology is in the source.