# Google Unveils Gemini 4 Carbon Frontier Model New AI model prioritizes advanced cybersecurity capabilities and long-horizon engineering task performance. - **Source:** [Hacker News](https://news.ycombinator.com) - **Published:** 2026-10-28 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/9c0d4c32-6fc4-49fa-ad76-ceb2da090c5c ## Launch of Gemini 4 Carbon Google DeepMind officially announced Gemini 4 Carbon on October 28, 2026, positioning it as their strongest frontier model to date. This release follows the launch of Gemini 4 Argon by only two weeks. The model is specifically engineered to excel in complex cybersecurity tasks and sustained, long-horizon agentic engineering workflows. ## Cybersecurity Performance Benchmarks Carbon ranks first on the Artificial Analysis Intelligence Index v5.0 with a score of 59. It demonstrates significant leadership in SEC-bench Pro at 88.2% for vulnerability detection, ExploitGym at 54.8% for exploitation, and CWE-bench at 71.0% for patching. These metrics highlight its specialized utility for modern, high-stakes cybersecurity defense operations. ## Agentic Workflow Capabilities Designed for complex, multi-step engineering, Carbon supports sustained agentic workflows that can persist for days. Internal testing shows the model successfully navigates dead ends to resolve intricate bugs within large codebases. This capability represents a shift toward AI agents that handle low-level system maintenance without requiring constant human intervention. ## Web Development Improvements Beyond security, Carbon shows substantial gains in full-stack web development and multimodal understanding. The model achieved a significant jump of 89 Elo points on the WebDev Arena, reaching a total score of 1768. This improvement reflects better output quality and artifact generation, making it a robust tool for professional software developers. ## Real-world Defensive Impact The Chrome team reports that Carbon performs ten times better than other frontier models when generating high-quality patches. Additionally, the cloud security platform Wiz observes that Carbon leads in web penetration testing benchmarks. These practical applications confirm the model's effectiveness in real-world scenarios, specifically regarding vulnerability remediation and secure code generation. ## Safety and Alignment Strategy Google employs a multi-layered safety stack, including real-time misuse defenses and prompt-injection resilience. Building on the alignment progress made with Argon, Carbon is specifically trained to handle high-pressure tasks appropriately. The company is currently engaging in the U.S. government’s voluntary pre-release testing process to ensure reliability before broader public deployment. ## Phased Deployment Approach Access is currently limited to a subset of cyber defenders, with plans to expand to trusted partners via the Fairwind Program. A wider release for developers, businesses, and consumers will follow after further iteration on guardrails. This deliberate rollout strategy aims to balance rapid innovation with necessary safety and security oversight. ## Pricing and Future Outlook Carbon launches with an introductory price of $2 per million input tokens and $10 per million output tokens. After the introductory period, rates will double to $4 and $20 respectively. Cached input tokens receive a 95% discount, providing a cost-effective path for enterprises integrating the model into long-term engineering pipelines. ## Sources covering this story - https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._