OpenAI Outpaces Anthropic: GPT-5.5-Cyber Surpasses Claude Mythos 5

OpenAI Takes the Lead Over Anthropic with GPT-5.5-Cyber

In recent weeks, the tech world has been buzzing about Claude Mythos, the artificial intelligence developed by Anthropic. This AI is recognized for its remarkable ability to identify security vulnerabilities. However, Claude Mythos 5 is currently suspended following a directive issued by the U.S. government, preventing its use for cybersecurity tasks. This situation has paved the way for a new player in the market.

OpenAI recently unveiled GPT-5.5-Cyber, a specialized version of its GPT-5.5 model optimized for cybersecurity applications. According to evaluations published by OpenAI, this new AI outperforms Claude Mythos 5 in this specific domain. Test results show that GPT-5.5-Cyber achieved an impressive score of 85.6% on the CyberGym benchmark, while Mythos 5 reached 83.8%. This benchmark serves as a reference for assessing the ability of AI models to analyze software vulnerabilities.

It is important to note that GPT-5.5-Cyber is not intended for the general public. Like Anthropic, OpenAI restricts access to this specialized AI to a limited number of organizations, which primarily use it for defensive purposes. This strategy aims to ensure that these powerful tools are not used maliciously.

Alongside the announcement of GPT-5.5-Cyber, OpenAI also revealed an update to Codex Security, its cybersecurity-focused plugin. Additionally, the company has expanded access to its Daybreak program. This program allows certain organizations to use Codex Security as well as GPT-5.5 with Trusted Access for Cyber, an AI specialized in cybersecurity but not at the same level as GPT-5.5-Cyber.

The timing of OpenAI's announcements appears strategic. Indeed, Anthropic is still unable to restore access to its Claude Mythos 5 and Claude Fable 5 models. This suspension follows a request from the U.S. government, which cited national security concerns to restrict access to these AIs for foreign nationals. In response, Anthropic chose to suspend access to these models for all users.

Discussions are currently underway between Anthropic and the U.S. government to attempt to lift these restrictions. Recently, after a meeting with Anthropic's CEO, Donald Trump stated that he no longer considers this AI lab a threat to national security, which could pave the way for restoring access to Anthropic's AIs in the near future.

With the introduction of GPT-5.5-Cyber, OpenAI seems to be gaining an edge in the field of cybersecurity, capitalizing on the current difficulties faced by Anthropic. This situation highlights the intense competition in the specialized artificial intelligence sector, where each technological advancement can quickly change the game.