Gemini 4 Carbon Opens to Paid Users at $2 per Million Tokens

Key takeaways

• Google began rolling out Gemini 4 Carbon on November 10, 2026, to paying developers and enterprise customers, with consumer access for Google AI Ultra subscribers.

• Introductory API pricing is $2 per million input tokens and $10 per million output tokens through November 30. The listed rates rise to $4 and $20 after that period.

• Google's /swarm agent-team feature is a preview in Antigravity. Carbon can use more tokens than Gemini 4 Argon, so a lower token price does not guarantee a lower cost per completed task.

Google opened Gemini 4 Carbon to paying developers and enterprise customers on November 10, 2026, and began rolling it out to Google AI Ultra subscribers in the Gemini app, Search and Antigravity. The model targets difficult, multi-step work, with introductory API rates of $2 per million input tokens and $10 per million output tokens.

What happened with Gemini 4 Carbon?

Google announced the wider release in a November 10 post from Google DeepMind senior vice president Koray Kavukcuoglu. Carbon had first been made available to a limited group of cybersecurity defenders through Google's Fairwind program about two weeks earlier. The new rollout expands access to paying developers, businesses and selected consumer subscribers.

Developers can use Carbon through the paid Gemini API tier in Google AI Studio. Enterprise customers can access it through Gemini for business. For consumers, Google is rolling it out to AI Ultra subscribers in the Gemini app, AI Mode in Search and Google Antigravity.

Google positions Carbon for sustained reasoning: complex coding and debugging, long-running software projects, cybersecurity tasks and scientific research. The company says Gemini 4 Argon remains the better fit for most software engineering and knowledge work, while Carbon is intended for harder jobs that need more reasoning and verification.

The release also introduces a preview of /swarm in Antigravity. Google says a developer can give one prompt to a coordinated group of AI agents, assign different roles and track their work across parallel workspaces. The agents can check one another's changes. This is a preview feature, not a promise that every Gemini user has access to a fully available multi-agent system.

Google says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench. These are company-reported benchmark claims, not a guarantee of results on every workload. Read Google's availability announcement and The Verge's rollout report.

Why the release matters

Analysis: Carbon's public rollout moves it from a restricted cybersecurity preview into paid developer and enterprise products, while also bringing it to Google's highest consumer subscription tier. It gives teams another option for demanding work, but it is not a free replacement for the Gemini experience most people already use.

During the introductory period, one million input tokens cost $2 and one million output tokens cost $10. A workload using one million of each would cost $12 at those rates. After November 30, the same token volumes would cost $24 at the announced $4 input and $20 output rates.

That calculation is not a forecast of a real project's bill. Google says Carbon uses more tokens than Argon to reason through difficult tasks and run longer verification loops. Independent coverage also cautions that retries and corrections can change the cost of finishing a job. VentureBeat's pricing analysis examines the difference between token rates and task costs.

This is a different announcement from Big Zero's coverage of Google's new TPU systems, which are AI chips, and Anthropic's API price changes. Carbon is a hosted model; this update concerns who can use it, what the API costs and how Google plans to support agent workflows.

What Gemini users and developers should do now

If you build with the API: Check that your project is on the paid Gemini API tier and confirm Carbon is available in your account and region. Start with a contained evaluation rather than routing every production request to the new model.

If you run an engineering team: Compare Carbon with Argon on the same representative tasks. Record input and output tokens, retries, latency, successful completion rate and human corrections. The useful metric is cost per accepted result, not just cost per million tokens.

If you are an enterprise customer: Confirm which Gemini for business or Antigravity access path your organization has enabled. Review data-handling rules, connected-app permissions and human approval steps before agents can modify code, documents or business systems.

If you use the Gemini app: The consumer rollout is for Google AI Ultra subscribers. Google has not announced a date for lower subscription tiers or free users. If Carbon does not appear in your account, the public announcement does not establish that your plan is eligible.

A practical starting point:

For another model-pricing example, see Big Zero's Gemini Omni 1.2 Flash breakdown. It covers a different product, but the same buyer lesson applies: compare the cost of the finished output, not a single headline rate.

What is still unknown and what to watch next

Google has not given a public date for Carbon access on lower-priced Gemini plans or the free tier. The company also describes /swarm as a preview, so its final availability, controls and limits may change.

Google has published its own benchmark results, but teams should test Carbon on their codebases, documents and evaluation sets before treating those claims as a purchasing decision. The same applies to cost: a longer reasoning process can consume more tokens even when the per-token price is lower.

The first pricing date to watch is December 1, 2026, when the announced introductory rates are scheduled to end. Developers should also watch Google's API documentation and account rollout notices for regional availability, rate limits and changes to Antigravity.

FAQ

Is Gemini 4 Carbon free?

Google's November 10 rollout lists paid Gemini API access for developers and enterprises, and Google AI Ultra access for consumers. Google has not announced when or whether the model will reach free Gemini users.

How much does Gemini 4 Carbon cost?

The introductory API rates are $2 per million input tokens and $10 per million output tokens through November 30, 2026. Google says the rates will become $4 and $20, respectively, after the introductory period.

What is the /swarm feature?

It is a preview in Google Antigravity that lets a developer coordinate multiple AI agents from one prompt. Google says the agents can take different roles and work in parallel, with checks on their changes.

Is Carbon better than Gemini 4 Argon?

Google describes Carbon as its model for more difficult, long-running tasks, but says Argon remains the ideal choice for most software engineering and knowledge work. The best choice depends on the task, output quality and total cost after retries.

Bottom line

Gemini 4 Carbon gives paid developers, businesses and AI Ultra subscribers access to Google's model for demanding, multi-step work, with a temporary API price that rises after November 30. Test it against Argon on a real workload and measure the cost of a successful result before moving production traffic.