Key Notes
- Google has expanded Gemini 4 Carbon to paying API customers and is rolling it out to Google AI Ultra subscribers in Gemini, Antigravity and AI Mode in Search.
- Antigravity’s /swarm preview coordinates workers and verifier agents in parallel, while Google’s office reconstruction demo required 10 overnight loops after a single initial prompt.
- Standard API prices remain $2 per million input tokens and $10 per million output tokens through November 30, then double to $4 and $20 on December 1.
Google has opened broader access to Gemini 4 Carbon, its most capable AI model, alongside a preview of agent teams in Antigravity. The November 10 launch brings Carbon to paying API customers and begins its rollout to Google AI Ultra subscribers in Gemini, Antigravity and AI Mode in Search.
The company announced the expansion two weeks after Carbon’s initial release to selected cybersecurity defenders. As we reported at that unveiling, access was initially limited. The new launch makes the model available for a much broader range of development and knowledge work, while specialist Carbon for cyber access continues through the Fairwind Program.
Antigravity Turns One Prompt Into an Agent Team
The accompanying /swarm preview changes how developers assign work. Antigravity describes a coordinator that delegates tasks to parallel workers, with separate verifier agents running tests and challenging changes. Workers use isolated Git worktrees, allowing them to edit separate working copies, and communicate through a shared message board.
Teams can combine models rather than give every agent the same expensive assignment. Carbon can handle architectural decisions, coordination and auditing, while lighter models perform routine edits. A live board, agent roster and timeline let users inspect the work and intervene. The preview is rolling out in the Antigravity app and CLI; Carbon is also available to Ultra subscribers through its IDE extensions.
Google warns that parallel agents can consume quota substantially faster than a standard conversation, even though idle waits do not consume tokens. The attraction is concurrent implementation and checking, but the cost still depends on how much work the entire team performs.
Seven Planning PDFs Become an Interactive Office
Google’s official demonstration shows a Carbon team reconstructing its Platform 37 office from seven public planning PDFs. The source material contained 206 extracted engineering drawing views. According to Google, the team produced an interactive Three.js model with about 906,000 triangles across 10 overnight loops.
The example illustrates the difference between issuing one instruction and completing a task in one pass. The agents worked through repeated cycles to produce the reconstruction. It is a company-selected demonstration, and the published example does not establish the accuracy of every architectural detail or the performance users should expect on unrelated documents.
The Introductory API Price Ends November 30
Google’s pricing documentation lists standard API rates of $2 per million input tokens and $10 per million output tokens through November 30. On December 1, those rates rise to $4 and $20 respectively. Output billing includes thinking tokens, and Carbon is unavailable on the API’s free tier.
For scale, one million input tokens and 100,000 billed output tokens would cost $3 at the introductory standard rates and $6 from December, before caching discounts or separate tool charges. More agents and longer verification loops can increase the tokens required to finish a project, even when the price per token is unchanged.
The model documentation specifies support for text, images, video, audio and PDFs as inputs, with text output and a unified context limit of 1,048,576 tokens. That gives the agent workflows room to work with substantial source material, while leaving the practical choice of model and team size dependent on the task’s demands and budget.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
