AI Research

Google's Carbon hunts 0-days in two hours

DeepMind released Gemini 4 Carbon with top cyber scores, same pricing as Argon but deeper reasoning for defenders and builders.

StartupHub.ai Staff

By StartupHub.ai Staff, reviewed by Daniel Singer, Editor

2 min read

Engineers tracking multi-agent Gemini 4 Carbon cyber scan
Credit: Deepmind

Google DeepMind opened Gemini 4 Carbon to cyber defenders through its Fairwind Program. The release targets open-source code and SaaS infrastructure, with exploit work gated to vetted testers rather than open remote access.

Is Carbon just Argon with a new name? No. Carbon spends more tokens than Argon to explore, reason deeply and run extended verification loops, yet Google lists both at $2 per million input tokens and $10 per million output tokens as of this release, per DeepMind.

Koray Kavukcuoglu, SVP at Google DeepMind and chief AI architect, positions Carbon for tasks that test model limits like multi-step debugging and open-ended research. Carbon ranks No. 1 on CursorBench 4.0, Perplexity WANDR and Zapier AutomationBench at more than 2X lower cost per task than the next best external model, per DeepMind. For cyber it leads on SEC-bench Pro for finding flaws, ExploitGym for exploiting them and CWE-bench v1 for patching, and A 0-day is a flaw with no patch available when it is found.

Speed is the point.

In Project Fortify, Google’s autonomous pipeline from scanning to remediation, Carbon hit 100% recall on scanning and 95% precision on triage on an internal benchmark, plus a 25% relative recall gain over Argon on open-source repos. Google’s Cloud Vulnerability Research used Carbon to prove two open-source bugs were exploitable at PHP2Shell severity, with details held until disclosure completes. Wiz, as part of its Scan for Good program, used Carbon to find 0-days in two major SaaS providers in under two hours, work the team says usually takes weeks. Security leads now have to staff for verification, not just alerts, and assign Carbon to architecture, coordination and auditing in Google Antigravity with /swarm teams that Argon first pushed off local hardware while faster models run parallel work. Morphic still lists Argon as the model developers can use on its platform today and describes Carbon as the frontier tier above it, a sign third-party deployment lags Google AI Studio and Ultra.

Watch the pending disclosures to judge real severity, and watch whether Fairwind stays gated. Ultra subscribers can try Carbon now in Antigravity, the Gemini app and AI Mode in Search, while paying developers get API access in AI Studio and Gemini for business.

Sources

Get the free daily brief every morning

Every morning the day's AI news, new AI startups and new funding rounds. Free and takes a few minutes to read, unsubscribe at any time.

Join 64,000 other subscribers.

TopicsGoogleDeepMindGemini 4 CarbonGemini 4 ArgonWizAxiom BioProject FortifyAntigravity

StartupHub.ai Staff

Written by

StartupHub.ai Staff

Editorial team

The staff writers of StartupHub.ai, ranging from investment analysts to avid AI tool users, early adopters and critical enthusiasts. Backgrounds span engineering, business and the arts. We hold every piece to rigorous standards of research and review.

Daniel Singer

Written by

Daniel Singer

Editor, reviewed this article

Daniel Singer is an AI expert who has been writing about, covering and building AI technology since 2018.

Test your Agent Readiness now

Paste your website. How well ChatGPT, Claude and the other agents can read and use a site, in about a minute.

Watch

The videos behind the news, one flip at a time

The feed mixes the videos from our articles with startup demos and you flip through them like TV channels for free with no signup.