A LinkedIn post from Aikido Security describes internal benchmark testing of its Aikido Code Security Audit against leading AI cybersecurity models. According to the post, the company compared its tool with Claude Mythos and Codex Security powered by GPT-5.6 Sol on a code repository containing 89 known vulnerabilities.
The post suggests Aikido’s audit identified 68 vulnerabilities at a cost of $75, compared with 60 for Mythos at $157 and 58 for GPT-5.6 Sol at $125. The content highlights that Aikido’s approach uses smaller specialized models and reserves more expensive models for complex cases, which is presented as a driver of lower cost and higher detection rates.
According to the post, Aikido’s system detected additional high-severity issues, including a cross-tenant authentication flaw and an API key expiration mismatch, which were reportedly missed by Mythos. This framing positions Aikido’s AI Code Analysis as competitively differentiated in both efficacy and cost efficiency within AI-driven application security.
For investors, the benchmark results may indicate potential pricing power and margin advantages if the reported performance translates to broader customer environments. The emphasis on orchestration and model “harness” design over reliance on frontier models could also signal a scalable product strategy that leverages existing AI infrastructure while competing on security outcomes rather than raw model capabilities.