Google Debuts Gemini 4 Carbon, Its Leading Cybersecurity AI Model

WHO:

Mashable - Google announced Gemini 4 Carbon, its new frontier AI model.

MarkTechPost - Google DeepMind introduced Gemini 4 Carbon as a model for cyber defense and coding.

Techmeme - Koray Kavukcuoglu of Google described Gemini 4 Carbon as the company’s “strongest model with state-of-the-art cybersecurity” capabilities.

WHAT:

VentureBeat - Gemini 4 Carbon’s introductory pricing is scheduled to rise from $2 per million input tokens and $10 per million output tokens to $4 and $20, respectively.

The Verge - Gemini 4 Carbon ranked ahead of Claude Opus 5.5 and GPT-6 Astra on the Artificial Analysis Intelligence Index v5.0.

The Verge - Google said Carbon holds the number one spot on both SEC-bench Pro and ExploitGym; SEC-bench Pro measures discovery of real-world bugs, while ExploitGym measures the ability to turn vulnerabilities into exploits.

The Verge - Google said Carbon beat GPT-6 Astra on SEC-bench Pro.

The Verge - Google said Gemini 4 Carbon offers gains in coding and agentic tasks, and holds the top spot on the ExploitGym benchmark.

The Verge - Google said Carbon reduces unsupported claims by two times compared with Argon.

Ars Technica - Gemini 4 Carbon is essentially a slightly better version of Argon.

Mashable - Gemini 4 Carbon topped the Artificial Analysis Intelligence Index.

Mashable - Google announced Gemini 4 Carbon two weeks after Argon.

MarkTechPost - Gemini 4 Carbon scored 59 on the Artificial Analysis Intelligence Index and 88.2% on SEC-bench Pro.

Techmeme - Gemini 4 Carbon ranked #1 on Artificial Analysis’ new Intelligence Index.

Techmeme - Google characterized Gemini 4 Carbon as having “long horizon capabilities.”

HOW:

VentureBeat - Google is rolling out Gemini 4 Carbon to selected cybersecurity defenders through its Fairwind program.

The Verge - Google said Carbon’s cybersecurity improvements help it find vulnerabilities and generate secure code.

CNET - Access to Gemini 4 Carbon is restricted to selected cybersecurity experts.

The Verge - Google said its internal teams had seen Carbon run for days through dead ends while tackling complex, multi-step engineering tasks.

The Verge - Google said Carbon is better at avoiding shortcuts, resisting prompt injection attacks, and collaborating with users.

The Verge - Google plans to expand Carbon access beyond selected cybersecurity defense companies to other partners in its Fairwind Program.

MarkTechPost - Access to Gemini 4 Carbon starts with vetted cyber defenders.

MarkTechPost - Pricing is $2 per million input tokens and $10 per million output tokens.

(UPDATED October 29, 2026 10:59 UTC)

Google has unveiled Gemini 4 Carbon, a new AI model aimed at coding and cyber defense, and it took the top spot on the Artificial Analysis Intelligence Index, ahead of Claude Opus 5.5 and GPT-6 Astra. Google says Carbon leads on benchmarks for finding real-world software bugs and turning vulnerabilities into exploits, while improving coding and reducing unsupported claims compared with Argon. Access is currently limited to vetted cybersecurity defenders through Google’s Fairwind program, with introductory pricing of $2 per million input tokens and $10 per million output tokens.

(UPDATED October 29, 2026 03:00 UTC)

Google has introduced Gemini 4 Carbon, a frontier model focused on cybersecurity and coding that ranks first on the Artificial Analysis Intelligence Index, ahead of Claude Opus 5.5 and GPT-6 Astra. Google says Carbon also leads on benchmarks for finding real-world software bugs and turning vulnerabilities into exploits, while improving coding, long-running engineering work, and resistance to prompt injection; the company says it makes half as many unsupported claims as Argon. Access is initially limited to vetted cybersecurity experts, with Google planning to extend it to more partners in its Fairwind Program.

(UPDATED October 28, 2026 22:01 UTC)

Google has introduced Gemini 4 Carbon, a new AI model it describes as its strongest for cybersecurity, and says it ranks first on the Artificial Analysis Intelligence Index and the ExploitGym benchmark. Google says Carbon improves coding and long-running, multi-step tasks, resists prompt-injection attacks, and makes half as many unsupported claims as its predecessor, Argon. Access starts with vetted cyber defenders, with plans to expand to more Fairwind Program partners; pricing is $2 per million input tokens and $10 per million output tokens.

(UPDATED October 28, 2026 21:01 UTC)

Google has announced Gemini 4 Carbon, a new AI model for coding and cyber defense that the company calls its strongest yet on cybersecurity. It ranks first on the Artificial Analysis Intelligence Index, scoring 59, and achieved 88.2 percent on SEC-bench Pro; Google also describes it as having long-horizon capabilities. Access will initially be limited to vetted cyber defenders, with pricing set at $2 per million input tokens and $10 per million output tokens.

(UPDATED October 28, 2026 20:02 UTC)

Google has introduced Gemini 4 Carbon, a new AI model aimed at cyber defense and coding, which the company describes as its strongest model for cybersecurity. It ranked first on the Artificial Analysis Intelligence Index, scoring 59, and achieved 88.2 percent on SEC-bench Pro, with Google also touting its ability to handle long-running tasks. Access begins with vetted cyber defenders, and pricing is $2 per million input tokens and $10 per million output tokens.