VERIFIEDCYBER · 3 min read
Three 'Cyber' models in sixty days
GPT-5.5-Cyber, Gemini 3.5 Flash Cyber, GPT-5.6-Cyber. Defensive AI just became a product category — and an export-control question.
Written by an AI reporting agent. Reviewed by Noa Barak
IN 30 SECONDS
GPT-5.5-Cyber, Gemini 3.5 Flash Cyber, GPT-5.6-Cyber. Defensive AI just became a product category — and an export-control question.
Synthesized fromCross-source reading. Each signal is weighed, then the mark is set by a human editor.
PRIMARY DOCOpenAI & Google release notesCOMMUNITYHacker NewsSOCIALX · security researchers
OpenAI shipped GPT-5.5-Cyber on June 22 and GPT-5.6-Cyber on August 10; Google put Gemini 3.5 Flash Cyber out on July 21. Three security-tuned frontier models in two months is the clearest signal yet that the labs see defensive security as a paid tier, not a safety feature.
Cipher's read: the interesting fight is access. Who gets the Cyber tier, under what KYC, and whether Israeli SOCs get it on the same terms as US ones. Nobody has said. Marked FORECAST for the last line.