Google announces Gemini 4 Carbon, a frontier model focused on cybersecurity

Google announced Gemini 4 Carbon, a new frontier AI model aimed at software engineering and cybersecurity work. The company says it scored 88.2% on SEC-bench Pro and reports gains in vulnerability detection, exploitation and patching; it is initially making the model available to vetted cybersecurity defenders through its Fairwind program. Google says the model improves on its predecessor, Gemini 4 Argon, while a report says introductory pricing is $2 per million input tokens and $10 per million output tokens, rising later to $4 and $20.

AI-written summary, checked against the cited sources. How summaries work

Image: Ars Technica

Why it matters

Cybersecurity defenders selected for Google’s Fairwind program can access Gemini 4 Carbon for vulnerability-finding and remediation tasks. Buyers should note the reported introductory token rates of $2 per million input tokens and $10 per million output tokens, which are set to rise to $4 and $20 later.

What we know

Confirmed by several sources

Reported by one source

  • Google says Carbon achieved 54.8% on ExploitGym for exploitation. — The Hacker News
  • Google teams have seen Carbon spend days investigating and fixing bugs across large codebases and low-level systems. — Help Net Security
  • Google says Carbon is twice as likely as Argon to generate code free of vulnerabilities, while the Chrome team claims it is 10 times better at generating high-quality patches than non-Google models. — Ars Technica
  • Google says Carbon rose 89 points on WebDev Arena and ranks fourth on its leaderboard. — Ars Technica

Open questions

  • It is not yet clear when broader access to Gemini 4 Carbon will begin. — Techmeme, Ars Technica
  • The duration of the introductory pricing period is not specified. — The Register

Coverage 8 publishers

  1. TechmemeFirst report

    Gemini 4 Carbon: our next frontier model

    Gemini 4 Carbon is our strongest model with state-of-the-art cybersecurity and long horizon capabilities.

  2. Hacker News

    Gemini 4 Carbon: our next frontier model

    Gemini 4 Carbon is our strongest model with state-of-the-art cybersecurity and long horizon capabilities.

  3. Ars Technica

    Google announces Gemini 4 Carbon, its latest frontier AI model that you can't use

    It's almost like Google's got something to prove after taking a break from releasing new frontier AI models earlier this year. Just two weeks after rolling out Gemini 4 Argon, Google has announced Gemini 4 Carbon. This isn't the Flash model…

  4. The Register

    Google's Gemini 4 Carbon looks like a diamond in the rough

    Just two weeks after launching Gemini 4 Argon, Google has announced a better-performing version of the model that it will sell for the same price, at least temporarily. Koray Kavukcuoglu, SVP at Google DeepMind and chief AI architect at…

  5. Engadget

    Google's Gemini 4 Carbon promises 'significant leap' in AI coding and cybersecurity

    Google's latest Gemini 4 Carbon AI model boasts top benchmark scores in areas like cybersecurity and coding.

  6. Help Net Security

    Google says Gemini 4 Carbon beats its predecessors at finding and fixing security flaws

    Google’s Gemini 4 Carbon is a frontier AI model designed for software engineering and cybersecurity tasks, with safeguards built on those of its predecessor, Argon. SEC-bench Pro (Source: Google) Google says its teams have seen Carbon spend…

  7. The Hacker News

    Google's Gemini 4 Carbon Leads Benchmarks for Finding, Exploiting, and Patching Vulnerabilities

    Google on Wednesday took the wraps off Gemini 4 Carbon, its latest frontier model with state-of-the-art cybersecurity and long-horizon capabilities. "It shows significant gains across coding and agentic tasks as well as across key…

  8. SecurityWeek

    Google Touts Powerful Cybersecurity Capabilities in Gemini 4 Carbon Model

    Google’s new frontier model scored 88.2% on SEC-bench Pro and is rolling out first to vetted cyber defenders via its Fairwind Program.

Earliest report first. Articles stay on their publishers’ sites; each link opens the original.