Gemini 4 Carbon

Google DeepMind made Gemini 4 Carbon generally available on November 10, 2026, opening the model to developers, enterprises, scientists, and consumers through the Gemini API and AI Studio for paying accounts. Carbon is positioned as the strongest frontier model for cybersecurity capability and long-horizon agentic workflows across complex knowledge work, software engineering, and scientific reasoning. The launch post positions Carbon as the harder-end companion to Gemini 4 Argon, which remains the better fit for most software engineering and knowledge work, while Carbon uses more tokens for extended reasoning and verification loops. Carbon ranks first on CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench, with more than 2X lower cost per task than the next best external model, and Google is previewing a /swarm mode in Antigravity that turns one prompt into a tracked team of agents.

Google announced Gemini 4 Carbon on October 28, 2026 as its next frontier model, built for difficult long-horizon engineering problems with strong cybersecurity defense capability. Carbon ranks first on the updated v5.0 Artificial Analysis Intelligence Index at 59, and Google reported a firm lead on SEC-bench Pro at 88.2% for finding vulnerabilities, ExploitGym at 54.8% for exploitation, and CWE-bench at 71.0% for patching. At launch, access was staged through the Fairwind Program to a subset of cyber defenders and trusted partners ahead of general developer availability, with Google saying it was engaging in the U.S. government's voluntary pre-release model testing process. Carbon was priced to launch at the same introductory rate as Argon, $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95%.

The Google DeepMind model card for Gemini 4 Carbon, published November 10, 2026, describes the model as Google's frontier performer for long-horizon agentic workflows across complex knowledge work, engineering, cybersecurity, and scientific reasoning. It accepts text, image, audio, and video inputs and returns text output, with a combined input-output token context window of up to 1M tokens. Architecture, training dataset, training data processing, hardware, and sustainability details for Carbon are explicitly delegated to the Gemini 4 Argon model card, indicating Carbon builds directly on Argon. Distribution is listed across the Gemini App, Gemini API, Google AI Studio, Gemini for business, Google Antigravity, Google AI Mode, and the Fairwind Program for verified cybersecurity defenders, with downstream API access subject to Google's terms of use.