قوقل تعلن Gemini 4 Carbon: أقوى نماذجها في البرمجة والأمن السيبراني، ويبدأ مع جهات الدفاع السيبراني بسGoogle announces Gemini 4 Carbon, its strongest model for coding and cybersecurity, starting with cyber defenders only

Google has announced Gemini 4 Carbon, which it calls its strongest model, with its biggest strengths in cybersecurity and long tasks. The announcement comes two weeks after the launch of Gemini 4 Argon, according to Google's blog.

According to Google, Carbon ranks first on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59. On cybersecurity benchmarks it scores 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym and 71.0% on CWE-bench for patching. Google says its teams have seen it run for days on complex engineering tasks, pushing through dead ends to find the root cause of a bug and fix it.

Google also says Carbon cuts unsupported claims by about half compared with Argon, that instead of guessing it looks up the relevant sources and flags discrepancies before acting, and that it resisted every attack attempt in Gray Swan's indirect prompt injection benchmark. All of these figures come from Google itself.

The model is not available to everyone yet: it starts this week with a subset of cyber defenders through the Fairwind Program, then more trusted partners, and later developers, businesses and consumers. The introductory developer price is $2 per million input tokens and $10 per million output tokens.