Recent evaluations by the UK's AI Security Institute (AISI) show that OpenAI's GPT-5.5, which launched publicly last week, matches Anthropic's Mythos Preview in cybersecurity performance. The AISI tested both models on 95 Capture the Flag challenges, focusing on skills like reverse engineering and cryptography. GPT-5.5 achieved an average score of 71.4% on expert tasks, slightly surpassing Mythos Preview's 68.6%. Notably, GPT-5.5 completed a challenging disassembler task in 10 minutes and 22 seconds with a cost of $1.73 in API calls. In a simulated data extraction attack, GPT-5.5 succeeded in 3 out of 10 attempts, compared to 2 out of 10 for Mythos Preview. However, both models failed to perform well in the Cooling Tower simulation, which tests the disruption of power plant control software, a challenge that has stumped previous AI models as well.