Google Unveils Gemini 4 Carbon AI Model
Google launched Gemini 4 Carbon, a new frontier AI model focused on cybersecurity.
"Google just dropped Gemini 4 Carbon, a cybersecurity AI powerhouse. It's top-ranked but only for a select few. Get ready for smarter cyber defense, eventually."
Google announced Gemini 4 Carbon, its latest frontier AI model, on Wednesday. The company highlights "significant gains" in coding, agentic tasks, and cybersecurity with this release. Gemini 4 Carbon holds the top position on the Artificial Analysis Intelligence Index (v5.0) benchmark, surpassing Claude Opus 5.5 and GPT-6 Astra. It is also ranked number one on SEC-bench Pro and ExploitGym, benchmarks measuring AI's ability to find vulnerabilities and create exploits.
The new model offers improvements over Google's previous Argon model, notably cutting "unsupported claims" by two times. Gemini 4 Carbon is currently available to a select group of cybersecurity defense companies, with plans to expand to other partners in its Fairwind Program. Google states the model is designed for longer tasks, with internal teams observing it run for days to complete complex, multi-step engineering tasks.
Google emphasizes Carbon's enhanced capabilities in avoiding shortcuts, resisting prompt injection attacks, and collaborating with users. The model will launch at the same introductory price as Argon: $2 per million input tokens and $10 per million output tokens. Its focus on cyberdefense marks a "significant leap" in finding vulnerabilities and generating secure code.
This new AI model could significantly impact enterprise cybersecurity strategies by offering advanced tools for vulnerability detection and secure code generation. Businesses in defense and other sectors may gain access to more robust AI for protecting their digital assets and streamlining complex engineering tasks.
Relevant tools
Find the right AI tool for your business
Chat with Insta and get matched to the right tool in seconds.
Try Insta Tool Finder →