Full Breakdown
Google Unveils Gemini 4 Carbon, a Cybersecurity-Focused AI Model
By Drooid · · How we work
In this breakdown
01
Launch Overview
Google announced the release of Gemini 4 Carbon, the latest iteration of its Gemini series. The model is initially offered to a limited set of cybersecurity-defense firms and is slated for broader distribution through Google’s Fairwind Program. Pricing matches the introductory rate for the preceding Argon model—$2 per million input tokens and $10 per million output tokens.
02
Performance Benchmarks
Gemini 4 Carbon secured the top position on the Artificial Analysis Intelligence Index (v5.0) benchmark, outpacing Claude Opus 5.5 and GPT-6 Astra. It also leads the SEC-bench Pro benchmark, which evaluates an AI system’s ability to uncover real-world software bugs, and the ExploitGym benchmark, which measures the conversion of discovered vulnerabilities into functional exploits. In direct comparison, Gemini 4 Carbon surpassed GPT-6 Astra on the SEC-bench Pro metric.
Visual evidence
Images from the story
03
Technical Improvements
According to Google, the new model is engineered for extended, multi-step engineering tasks, with internal testing showing it can run for days without stalling on dead ends. It exhibits stronger resistance to prompt-injection attacks, reduces reliance on shortcuts, and collaborates more effectively with users. Transparency and honesty are emphasized: Gemini 4 Carbon reportedly cuts “unsupported claims” by half relative to Argon.
04
Anticipated Impact on Cyber Defense
Google frames Gemini 4 Carbon as a “significant leap” in cyber-defense capability, claiming it is better at locating software vulnerabilities and producing secure code. By improving vulnerability discovery and exploit mitigation, the model could raise the baseline security posture of partner organizations and influence broader industry standards for AI-assisted security testing.
05
Official Position from Google
Google states that Gemini 4 Carbon reflects the rapid pace of its frontier AI development and is intended to deliver more reliable, transparent assistance for complex cybersecurity challenges. The company highlights the model’s competitive pricing and its role in expanding AI-driven protection across the cyber-defense ecosystem.