OpenAI’s GPT-5.5 Cyber AI Tops CyberGym as Anthropic Models Remain Offline

OpenAI’s GPT-5.5-Cyber has topped the CyberGym leaderboard, according to source material from Decrypt. The ranking comes as Anthropic’s best-performing models are said to remain offline under a Trump administration export ban.

The development matters because cyber-focused AI benchmarks are becoming a more visible way to compare advanced model capabilities. For companies and markets watching the AI sector, leaderboard performance can shape perceptions of technical momentum, even when access restrictions or policy decisions affect which models are available for comparison.

Decrypt’s report frames the contrast around two major AI companies: OpenAI, whose GPT-5.5-Cyber is currently leading CyberGym, and Anthropic, whose top models are not active on the leaderboard because of the cited export ban. The source material does not provide pricing, release timing, technical scores, or details about the ban’s scope.

For crypto readers, the story sits at the intersection of AI capability, cybersecurity, and policy risk. The crypto ecosystem depends heavily on secure infrastructure, audits, wallets, exchanges, and smart contracts, making advances in cyber-oriented AI relevant even when the news is not directly about digital assets.

The key takeaway is narrow but notable: OpenAI’s cyber-focused model is leading the benchmark while a major competitor’s strongest models are sidelined. Without additional technical data, the ranking should be read as a benchmark result and availability snapshot, not as a complete measure of real-world security performance.