OpenAI launches GPT‑5.4‑Cyber
OpenAI released GPT‑5.4‑Cyber, a model tuned for defensive security tasks, and is distributing it via a tiered Trusted Access for Cyber programme rather than broad public rollout. The company says the model targets reverse engineering, vulnerability discovery and malware analysis for verified security professionals and enterprises. (reuters.com) (axios.com) (wired.com)
OpenAI has launched GPT-5.4-Cyber, a security-focused version of its flagship model, and is limiting access to vetted defenders instead of the general public. (openai.com) The company announced the release on April 14, 2026, saying the model is tuned for defensive work such as reverse engineering, vulnerability research, and malware analysis. Reuters reported the rollout comes first to vetted security vendors, organizations, and researchers. (openai.com) (reuters.com) OpenAI is also expanding its Trusted Access for Cyber program to thousands of verified individual defenders and hundreds of teams protecting critical software. Higher tiers require stronger identity checks, and the top tier can request access to GPT-5.4-Cyber. (openai.com) (reuters.com) Cybersecurity work here means finding weak spots in software before attackers do. OpenAI says it is making models more “cyber-permissive” for verified users, which means the system will answer more sensitive security questions than a standard public chatbot would. (openai.com) (axios.com) The timing follows Anthropic’s April 7 launch of Claude Mythos Preview under a restricted program called Project Glasswing. Anthropic said select organizations could use that model for defensive cybersecurity, and Reuters said the company claimed it had already found thousands of major vulnerabilities. (anthropic.com) (reuters.com) OpenAI is taking a different access approach from Anthropic, according to Axios. Instead of a small partner list, OpenAI described a tiered system meant to widen access while still screening who gets the most capable tools. (axios.com) (openai.com) The model sits on top of GPT-5.4, which OpenAI released on March 5, 2026, for ChatGPT, the application programming interface, and Codex. OpenAI said GPT-5.4 supports up to 1 million tokens of context and was built for coding, tool use, and long professional workflows. (openai.com) OpenAI has been building this security push for months. The company introduced Trusted Access for Cyber in February 2026, pledged $10 million in application programming interface credits for cyber defense, and in March put Codex Security into research preview for finding and patching software flaws. (openai.com 1) (openai.com 2) OpenAI says more capable models are coming “over the next few months,” and it is treating GPT-5.4-Cyber as an early test of how to release them with tighter controls. For now, the company is betting that verified defenders, not the public internet, are the right first users. (openai.com)
