Google opens Gemini 4 Carbon to developers, businesses and Ultra users

Google expanded access to Gemini 4 Carbon on 10 November 2026, bringing the model to paying developers, businesses and Google AI Ultra subscribers. Carbon was originally announced on 28 October with access limited to selected cybersecurity defenders through Google's Fairwind Program. The new release moves the model beyond that initial group and adds a preview of coordinated agent teams in Antigravity.

Developers can now use Carbon through the paid Gemini API in Google AI Studio and Android Studio. Companies can access it through Gemini for business. For Ultra subscribers, Google is rolling it out in the Gemini app, AI Mode in Search and Antigravity. The specialized Carbon for cyber access route continues through Fairwind, with applications reviewed separately.

The API's standard introductory rates are $2 per million input tokens and $10 per million output tokens through 30 November 2026. From 1 December, those rates become $4 and $20. Google's pricing documentation says output billing includes thinking tokens, and Carbon is unavailable on the free API tier. These are token rates; the cost of completing a project also depends on how much reasoning and repetition it requires.

Alongside the rollout, Google introduced /swarm in preview for the Antigravity app and command line. Its Antigravity announcement describes a coordinator, specialized workers in separate workspaces and verifiers that run tests and challenge changes. Teams can assign Carbon to architecture, coordination or review while using lighter models for routine edits. A shared board lets users follow the agents' activity.

Google cautions that these parallel teams consume usage quota much faster than an ordinary conversation. It also positions Argon as suitable for most engineering and knowledge work, with Carbon aimed at particularly demanding coding and research problems. For companies, that makes model choice an operational decision: compare completion quality, total consumption and the effort required to review the result.

The immediate significance is broader access to a model that previously required selection, combined with a new way to organize longer projects. Developers can now test it against their own workflows. Google's demonstrations and performance claims remain evidence presented by the supplier; the rollout itself does not establish that every team will finish work faster or spend less.