Skip to content
NewA calmer, faster ZeroHour

Google releases Gemini 4 Carbon, initially limited to cyber defense firms

Original title: Google launches Gemini 4 Carbon with a ‘significant leap’ in cybersecurity

The Verge · AIby Emma Roth ()1 min read1
Part of a story covered by 11 sourcesGoogle limits Gemini 4 Carbon to Fairwind cyber defendersMerged summary & timeline
AI summary · x-ai/grok-4.7Summaries by Vani AI

Google says Carbon tops two bug-finding benchmarks and costs the same as Argon, at $2 and $10 per million tokens.

Google released Gemini 4 Carbon, a frontier model it says improves on Argon in coding, long agentic tasks, and cybersecurity. Access is limited to selected cybersecurity defense companies, with a later expansion to other Fairwind Program partners. Google claims Carbon leads the Artificial Analysis Intelligence Index v5.0 and is stronger at finding bugs, generating secure code, and resisting prompt injection.

Unconfirmed: Benchmark rankings and security gains are Google's claims, not independently verified here.

Key facts
  • Google says Carbon ranks first on SEC-bench Pro and ExploitGym.
  • Claude Opus 5.5 and GPT-6 Astra trail it on the cited index, Google says.
  • Introductory pricing is $2 per million input tokens and $10 per million output.
  • Google says unsupported claims are cut twofold compared with Argon.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

1 min read at theverge.com

Google has released its latest frontier model, Gemini 4 Carbon, offering "significant gains" in coding, agentic tasks, and cybersecurity. In an announcement on Wednesday, Google highlights Carbon's top spot on the Artificial Analysis Intelligence Index (v5. 0) benchmark, with Claude Opus 5.

The Verge · AI

By Emma Roth · theverge.com

Read the full article at The Verge · AI

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.