> ## Content Index
> Fetch the complete content index at: https://www.skimwire.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Google's Gemini 4 Carbon tops an AI leaderboard, but only cyber defenders can use it for now
- URL: https://www.skimwire.com/google-gemini-4-carbon-cybersecurity-model/
- Published: 2026-10-28T19:41:33.000Z
- Updated: 2026-10-28T19:41:33.000Z
- Description: Google's new frontier model leads on finding, exploiting and patching software flaws. It goes to vetted cyber defenders first, with no public date.
- Author: Jeremy Moon
- Tags: AI, Security

## The skim

- Google announced Gemini 4 Carbon on Wednesday, two weeks after Gemini 4 Argon. It calls Carbon its strongest model yet, with a focus on cybersecurity and long, multi-step coding work.
- You can't use it yet. Google is giving it first to a subset of cyber defenders through its Fairwind Program and has set no date for developers or consumers.
- Google's numbers are strong but mostly self-reported: first place on the Artificial Analysis Intelligence Index at 59, and top scores on benchmarks for finding, exploiting and patching software flaws.

## What Google announced

Carbon "is our strongest model with state-of-the-art cybersecurity and long horizon capabilities," Google DeepMind's Koray Kavukcuoglu wrote in [the announcement](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/?ref=skimwire.com). Google says it ranks first on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59, and leads three security benchmarks: 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploiting them, and 71.0% on CWE-bench for patching them. Google notes those security scores were run without guardrails.

The rest is mostly internal. Google says Carbon more than doubles Argon's success rate at writing secure code, and that teams have seen it work for days on hard engineering tasks like tracking down and fixing subtle bugs in large codebases. The Chrome team reported Carbon did more than 10 times better than "another lab's large frontier model" at producing high-quality patches. Google didn't name the rival.

Google also leans on safety claims: zero successful attacks on Gray Swan's indirect prompt injection benchmark (Argon allowed 0.7%), successful cheating on Terminal-Bench 4.0 cut to 0.0%, and high-severity misalignment alerts down almost 80% in internal coding use.

## Why it matters

A model that's very good at finding and exploiting bugs is useful to defenders and attackers alike, so Google is holding Carbon back, just as it did with Argon. Google says it's taking part in the U.S. government's voluntary pre-release testing and will widen access gradually through Fairwind before opening Carbon to developers, businesses and consumers "as soon as possible."

Some caution is in order. Ars Technica's Ryan Whitwam [points out](https://arstechnica.com/ai/2026/10/google-announces-gemini-4-carbon-its-latest-frontier-ai-model-that-you-cant-use/?ref=skimwire.com) that Carbon beats Argon by slim margins across the board, and that "for all we know, it could be a post-training upgrade of Argon." Frontier models now score within a few points of each other, he notes, with Carbon edging out OpenAI's GPT-6 Astra and Anthropic's Claude Opus 5.5 in almost every test Google shared. Since almost nobody outside Google can use Carbon yet, these results can't easily be checked independently.

For businesses the price is known. Carbon will launch at Argon's introductory rate of $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 when the introductory period ends. Google hasn't said when that is.

## What's next

If Carbon follows Argon's path, wider access could come within weeks, Ars suggests, probably starting with Google AI Ultra subscribers and the Gemini API. Google gave no timeline. It's also a busy day for Alphabet, which reports third-quarter earnings after the market closes Wednesday.

*Sources:* [*Google*](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/?ref=skimwire.com)*,* [*Ars Technica*](https://arstechnica.com/ai/2026/10/google-announces-gemini-4-carbon-its-latest-frontier-ai-model-that-you-cant-use/?ref=skimwire.com) *and* [*Reuters*](https://www.tradingview.com/news/reuters.com,2026:newsml%5FL5N46E2P8:0-google-announces-gemini-4-carbon-priced-at-2-per-mln-input-tokens-10-per-mln-output-tokens/?ref=skimwire.com)*.*