Google launches Gemini 4 Carbon, its most powerful model yet, and brings it to AI Mode in Search
The new frontier model is rolling out via API to paying developers and businesses, and to Google AI Ultra subscribers in the Gemini app and in Search.

Google released Gemini 4 Carbon on November 10, describing it as the company's most robust model. It is available to developers, businesses, scientists and consumers. Google AI Ultra subscribers also get the model in the Gemini app and in AI Mode in Search.
The announcement was published by Google Brasil on November 10. According to the company, Carbon was built for tasks that push a model to its limits, such as complex, multistep coding and debugging and open-ended scientific research. To get there, it spends more tokens than the recently launched Gemini 4 Argon to explore, reason and check its own work. Argon remains the recommended choice for most software engineering and knowledge work tasks.
Where the model is available
Through the API, Carbon is available in Google AI Studio and Gemini Enterprise for all paying developers and businesses. Launch pricing matches Argon: $2 per million input tokens and $10 per million output tokens.
For Google AI Ultra subscribers, the rollout covers Google Antigravity, the Gemini app and AI Mode in Search. Two weeks earlier, the model had already been opened to a group of cybersecurity professionals through the Fairwind Program.
Performance and cost
Google says Carbon ranks 1st on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench, at a cost per task more than 2x lower than the next-best external model. On the Multimodal Knowledge Agent (MKA) benchmark, built with Artificial Analysis to measure how agents handle files packed with charts, tables and floor plans, the model placed 2nd, with an average cost per task more than 6x lower than the top-ranked model.
The company also introduced /swarm in Google Antigravity, a feature that turns a single prompt into a coordinated team of agents you can follow in real time. Carbon can take on the roles that call for deeper reasoning, such as architectural design, coordination and auditing, working alongside faster, lighter models.
According to the announcement, the model follows strict schemas and formatting to produce spreadsheets, presentations and reports. It also synthesizes information from large sets of documents and data in legal, financial and operational fields.
What this means for brand teams
Carbon is now available in AI Mode in Search and in the Gemini app, for now only to Google AI Ultra subscribers. The announcement does not detail any changes to how AI Mode cites sources or surfaces brands in its answers.
Key questions
Who can use Gemini 4 Carbon?
Paying developers and businesses, via the API in Google AI Studio and Gemini Enterprise. Google AI Ultra subscribers get the model in Google Antigravity, the Gemini app and AI Mode in Search.
How much does Gemini 4 Carbon cost in the API?
Launch pricing is the same as Gemini 4 Argon: $2 per million input tokens and $10 per million output tokens.
What is the difference between Carbon and Argon?
Argon remains the recommended model for most software engineering and knowledge work tasks. Carbon was built for more demanding tasks and uses more tokens to reason and check its own work.
Source: Google Brasil.