Skip to content
News briefAI-2026-0231

Gemini 4 Carbon opens to all paying developers

Google DeepMind's strongest model left its cyber-defender preview and went on general release at Argon's $2 and $10 per million tokens.

2 minMain AI HubFresh · 10 Nov

Google DeepMind made Gemini 4 Carbon generally available on 10 November. The company calls it its strongest model, aimed at long-horizon agent work in software engineering, knowledge work and scientific reasoning, and at cybersecurity. Two weeks earlier Carbon had gone only to a group of cyber defenders through Google's Fairwind Program; it is now open to developers, enterprises, scientists and consumers.

Google positions it above Gemini 4 Argon, which it still recommends for most software and knowledge work. Carbon is meant for the hardest tasks, such as multi-step coding and debugging or open-ended research, and it deliberately spends more tokens than Argon to explore, reason and run longer verification loops. For now there is no premium for that: the API costs the same introductory $2 per million input tokens and $10 per million output tokens as Argon, for all paying developers and enterprises in Google AI Studio and Gemini for business. It is also rolling out to Google AI Ultra subscribers in Antigravity, the Gemini app and AI Mode in Search.

The benchmark claims lean on third-party tests built around real workflows. Google says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench at more than 2x lower cost per task than the next best external model. On a new Multimodal Knowledge Agent benchmark, built with Artificial Analysis to test reasoning over dense files with charts, drawings and tables, Carbon ranks second while its average cost per task is more than 6x lower than the leader's.

The release comes with a preview of /swarm in Google Antigravity, which turns a single prompt into a team of agents that can be tracked as they work. Google's suggested split gives Carbon the roles that need the most reasoning, such as architecture, coordination and auditing, while lighter models work in parallel isolated workspaces and adversarial verifiers test every change. One demonstration rebuilt a Google office as a 906,000-triangle interactive 3D model from seven public planning PDFs over ten overnight loops.

Retold from Google DeepMind. This is a summary in our own words; follow the link for the original reporting.

Questions

How much does Gemini 4 Carbon cost?
Google DeepMind lists the same introductory API price as Gemini 4 Argon: $2 per million input tokens and $10 per million output tokens, for paying developers and enterprises in Google AI Studio and Gemini for business.
What is /swarm in Google Antigravity?
A preview feature that turns one prompt into a coordinated team of agents. Google suggests giving Carbon the roles that need the deepest reasoning, such as design, coordination and auditing, and pairing it with lighter models and adversarial verifiers.

Read next

Across the network

Desks that share a zone with this one on the BITBRIEF coverage map.

Terms defined