Google is rolling out its latest AI model, Gemini 4 Carbon, to the general public after introducing it to a select group of “cybersecurity advocates” two weeks ago. Google says Carbon is its “most powerful model” to date and is capable of competing with the best systems from Anthropic and OpenAI.
Google says the Gemini 4 Carbon “pushes the frontier even further” than the Argon, referring to the company’s other frontier model launched just last month. The company said Argon will likely remain “the ideal choice for the majority of software and knowledge engineering work,” but that Carbon is “designed for tasks that test the limits of a model’s intelligence,” such as complex, multi-step coding and open scientific research.
To highlight Carbon’s usefulness for long-term work involving AI agents, Google said it is also planning a “swarm” feature on its Antigravity agent development platform, which will allow users to turn a single prompt “into a coordinated team of agents.”
Carbon is rolling out today via the Gemini API in Google AI Studio and Gemini for business to all paying developers and businesses, Google said in a blog post. It will also be available to all Google AI Ultra subscribers in the Gemini app, AI Mode in Search and on Antigravity. It’s unclear when – or if – Carbon will be available to lower-tier subscribers or free users.
As with the Argon launch, Google sets a very competitive price for carbon. The model is launching with an introductory price of $2 per million input tokens and $10 per million output tokens, about a fifth of the cost of OpenAI’s GPT-6 Astra and a quarter of the cost of Anthropic’s Fable 5.2 model, which launched yesterday. Even after these costs double at the end of November – when the period for introducing lower prices ends – carbon will remain significantly cheaper than its border competitors on a per-token basis, although this does not necessarily translate into lower costs as models can use different numbers of tokens for the same tasks.
Benchmarks cited by Google during Carbon’s initial announcement suggest the model is one of the most powerful on the market for most metrics, behind Fable 5.2. It’s worth noting that companies can and do cherry-pick benchmark results to present their tools in the best possible light.
Gn tech
