Models & Research

Google makes Gemini 4 Carbon available to developers, enterprises and AI Ultra users

Gemini 4 Carbon, Google's cyber-focused frontier model, is now in the Gemini API, Gemini for business, Antigravity, the Gemini app and AI Mode, at Argon's introductory price until 30 November.

Gloved hands fitting liquid cooling tubes to GPU server boards, no logos or text
Source

Google said on 10 November that Gemini 4 Carbon is now available to developers, enterprises, scientists and consumers. In a Google blog post, Koray Kavukcuoglu, Google DeepMind’s chief AI architect, described Carbon as the company’s strongest model for cybersecurity work and for long-horizon agentic workflows in knowledge work, software engineering and scientific reasoning.

Carbon had a two-week preview for cybersecurity defenders in Google’s Fairwind Program. The post positions Gemini 4 Argon as the usual choice for most software engineering and knowledge work, and Carbon as the model for harder, multi-step coding, debugging and open-ended scientific research. Google says Carbon uses more tokens than Argon to explore thoroughly, reason deeply and run extended verification loops.

The company says Carbon ranks first on CursorBench 4.0, Perplexity’s WANDR and Zapier’s AutomationBench, at more than twice lower cost per task than the next best external model in those tests. Those rankings are Google’s account of independent benchmarks; they have not been re-run here.

Demis Hassabis, Google DeepMind’s chief executive, reposted a Cursor announcement on X that evening: “Gemini 4 Carbon is now available in Cursor! At high effort, it’s the top model on CursorBench at 60.4%.” The 60.4% figure appears in that Cursor post as relayed by Hassabis, not in the Google blog’s summary of CursorBench 4.0.

Google is also previewing /swarm in Google Antigravity, which it says turns a single prompt into a coordinated team of agents that can be tracked in real time. Developers can assign Carbon to roles that need deeper reasoning and pair it with faster models and adversarial verifiers.

On pricing, Carbon is on the Gemini API paid tier in Google AI Studio and in Gemini for business at the same introductory price as Argon: $2 per million input tokens and $10 per million output tokens. After 30 November 2026, the post says the price becomes $4 and $20 per million tokens for the public Carbon model. It is also rolling out to Google AI Ultra subscribers in Antigravity, the Gemini app and AI Mode in Search.

The post highlights cybersecurity results from early testers, including Project Fortify and work by Google’s Cloud Vulnerability Research team and Wiz’s Scan for Good program. It also reports safety figures, including a 0% prompt-injection attack success rate on Gray Swan’s IPI benchmark during the earlier Carbon announcement, while noting that robustness against more sophisticated attacks remains an area for improvement.

Scientists already in, or applying to, Google’s programme that offers Antigravity access will also be able to use Carbon, the post says, subject to eligibility, capacity and selection. A separate Gemini 4 Carbon Safety and Security report is linked from the announcement.

This report is not legal, financial, or health advice. See the disclaimer.

Primary sources

  1. Gemini 4 Carbon is now available Google
  2. Demis Hassabis on X X

Tags