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

DevFeed: [Google's Gemini 4 Carbon Leads Benchmarks for Finding, Exploiting, and Patching Vulnerabilities](<https://devfeed.tech/articles/google-s-gemini-4-carbon-leads-benchmarks-for-finding-exploiting-and-patching-vulnerabilities-106439.md>)

Original publisher: [Read original article](<https://thehackernews.com/2026/10/googles-gemini-4-carbon-leads.html>)

Author: The Hacker News

Published: 2026-10-29T13:42:40Z

Content type: news

Language: en

Sources: [Cyber Newsroom](<https://devfeed.tech/sources/cyber-newsroom.md>)

Topics: [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [code security](<https://devfeed.tech/topics/code-security.md>), [cve remediation](<https://devfeed.tech/topics/cve-remediation.md>), [vulnerability management](<https://devfeed.tech/topics/vulnerability-management.md>), [aws inspector](<https://devfeed.tech/topics/aws-inspector.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [capabilities](<https://devfeed.tech/tags/capabilities.md>), [carbon](<https://devfeed.tech/tags/carbon.md>), [coding](<https://devfeed.tech/tags/coding.md>), [cve](<https://devfeed.tech/tags/cve.md>), [cybersecurity](<https://devfeed.tech/tags/cybersecurity.md>), [exploit](<https://devfeed.tech/tags/exploit.md>), [frontier](<https://devfeed.tech/tags/frontier.md>), [frontier-model](<https://devfeed.tech/tags/frontier-model.md>), [gemini](<https://devfeed.tech/tags/gemini.md>), [gemini-4-carbon](<https://devfeed.tech/tags/gemini-4-carbon.md>), [google](<https://devfeed.tech/tags/google.md>), [google-s](<https://devfeed.tech/tags/google-s.md>), [latest](<https://devfeed.tech/tags/latest.md>), [model](<https://devfeed.tech/tags/model.md>), [patching](<https://devfeed.tech/tags/patching.md>), [shows](<https://devfeed.tech/tags/shows.md>), [software-security](<https://devfeed.tech/tags/software-security.md>), [this](<https://devfeed.tech/tags/this.md>), [vulnerabilities](<https://devfeed.tech/tags/vulnerabilities.md>), [vulnerability](<https://devfeed.tech/tags/vulnerability.md>), [with](<https://devfeed.tech/tags/with.md>)

## AI overview

Google says Gemini 4 Carbon leads the cited cybersecurity benchmarks for vulnerability discovery (SEC-bench Pro, 88.2%), exploitation (ExploitGym, 54.8%), and patching (CWE-bench, 71.0%); it also reports Carbon is twice as likely as Argon to generate secure code. Google says the model asks for human input when complex tasks are ambiguous and resisted all indirect prompt-injection attempts in its evaluations. Access is currently limited to some cyber defenders, with plans to expand to trusted Fairwind Program partners. The article also reports Anthropic's internal model found serious vulnerabilities in widely used software, which Google, Apple, and Microsoft have since fixed.

## Source excerpt

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 agent ... Read more Published Date: Oct 29, 2026 (2 hours, 47 minutes ago) Vulnerabilities has been mentioned in this article. CVE-2026-33096