Gemini 4 Carbon is now available

Google has unveiled Gemini 4 Carbon, the company's latest AI model, following its recent release of Argon to a select group of cybersecurity professionals. According to Google, Carbon represents the strongest model they have developed yet, capable of competing with industry leaders Anthropic and OpenAI. The company states that Carbon "pushes the frontier even further" beyond Argon, pointing to the company's previous frontier model that launched last month.

While Argon is likely the best option for most software engineering and knowledge work tasks, Carbon is tailored for more demanding, complex, and open-ended scientific research involving AI agents.

Google is previewing a "swarm" feature on its agentic development platform, Antigravity, which allows users to transform a single prompt into a coordinated team of agents. This new feature highlights Carbon's potential for long-horizon work that challenges the limits of a model's intelligence. The AI model is now available to all paying developers and enterprises via the Gemini API in Google AI Studio and Gemini for Business, as well as to all Google AI Ultra subscribers through the Gemini app, AI Mode in Search, and on Antigravity.

Pricing for Carbon is competitive, with an introductory rate of $2 per million input tokens and $10 per million output tokens. This is roughly a fifth of the cost of OpenAI's GPT-6 Astra and a quarter of Anthropic's Fable 5.2 model, which launched yesterday. Even after the introductory period ends in late November, Carbon will still be significantly cheaper than its competitors on a per-token basis.

However, this does not necessarily mean lower overall costs, as models may employ varying token usage for similar tasks. Google's benchmark results suggest that Carbon is among the most powerful models on the market, although it is important to note that companies may selectively present benchmark results to showcase their tools favorably.

This launch comes as Google seeks to regain its position at the forefront of AI research, following a period in which Anthropic and OpenAI appeared to dominate the frontier in AI development.