Skip to content
ShiftDelete.Net
Search news/ TR Türkçe
News Feed AI Smartphones Games Cars Computers Social Media Space Reviews Latest news
  1. Home
  2. AI

Gemini 4 Carbon Pricing Starts at $2 Before Doubling on Dec. 1

Google opened Gemini 4 Carbon to paying developers and AI Ultra users on Nov. 10. The $2 input rate ends Nov. 30, and reasoning tokens count on every bill.

Hakkı ALKANHakkı ALKAN

Published 4 min read

Glowing carbon crystal lattice above a server rack with streams of light, illustrating Gemini 4 Carbon pricing and access

Gemini 4 Carbon pricing starts at $2 per million input tokens and $10 per million output tokens, Google said on Tuesday, Nov. 10, 2026, as it opened its most capable model to all paying developers, enterprises and Google AI Ultra subscribers. That introductory rate ends on Nov. 30, and from Dec. 1 the price doubles to $4 and $20 per million tokens.

Two weeks ago, Carbon was limited to a small group of security teams in Google's Fairwind Program, as we reported when Google first unveiled Gemini 4 Carbon. Now it is a regular product, and the price is the part that matters most for anyone building with it.

Glowing carbon crystal lattice above a server rack with streams of light, illustrating Gemini 4 Carbon pricing and access

Who can use Gemini 4 Carbon now?

In a Google blog post, Koray Kavukcuoglu, SVP of Google DeepMind and Google's Chief AI Architect, said Carbon is available through the Gemini API in Google AI Studio and Gemini for business. That covers paying customers only, so the Gemini API free tier does not include it.

Consumers need Google AI Ultra, Google's top paid plan. Ultra subscribers are getting Carbon in the Gemini app, in AI Mode in Search and in Antigravity, Google's agent-based coding tool. In the Gemini app, Carbon appears in the model picker as the option for "Most complex work," while Gemini 4 Argon stays available from AI Pro upward, according to 9to5Google.

What changes on Dec. 1?

The introductory Gemini 4 Carbon pricing matches what Google charged at the launch of Gemini 4 Argon. After Nov. 30, input costs $4 and output costs $20 per million tokens. Tokens are the small chunks of text a model reads and writes, and the reasoning tokens Carbon uses while it "thinks" are billed too.

That is the catch. Google says Carbon uses more tokens than Argon to explore a problem and check its work. So the same job can cost more on Carbon even at an identical per-token price. Google itself still recommends Argon for most coding and office work, and positions Carbon for hard multi-step debugging and open-ended research.

How does Gemini 4 Carbon pricing compare with rivals?

Per token, Carbon undercuts its closest rivals. The Verge put the introductory price at roughly a fifth of OpenAI's GPT-6 Astra and a quarter of Anthropic's Claude Fable 5.2, which launched a day earlier. Even at the December price, Carbon stays at half of Fable 5.2's list rate.

Google also claims 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 outside model. On the new Multimodal Knowledge Agent benchmark built with Artificial Analysis, Google says Carbon placed second at more than 6x lower cost per task than the leader. These are vendor-selected results, and independent tests will matter more.

What is the new /swarm feature?

Alongside the rollout, Google is previewing /swarm in Antigravity. A single prompt turns into a coordinated team of agents that you can watch in real time. Carbon can take the roles that need the deepest reasoning, such as architecture, coordination and auditing, while lighter models handle routine work in separate workspaces and "adversarial verifiers" test every change.

Running many agents in parallel also burns through usage quota much faster than a normal chat, so teams should set limits before trying it on large projects.

Is Carbon safer than Argon?

Google reports that in simulated cyberattack tests Carbon blocked 99% of attack attempts, and its per-attempt breach rate fell to 0.8% from 3.1% on Argon. Google also cites a 0% success rate for attacks on the Gray Swan indirect prompt injection benchmark, while admitting that robustness against such attacks still needs work.

The model has already been used in real research. Google researchers used Carbon in AlphaProof Code, the framework that formally verified code behind Signal's encryption protocol.

Where you can use it

  • Developers: Gemini API paid tier in AI Studio and Android Studio.
  • Businesses: Gemini for business via API or model menu, Antigravity and CodeMender.
  • Consumers: Gemini app and AI Mode in Search, Google AI Ultra only.
  • Scientists: a free six-month Antigravity program, now covering Argon and Carbon, by application.
  • Security teams: Carbon for cyber through the Fairwind Program, by application.

What it means for you

If you build on the Gemini API, the next three weeks are the cheapest time to benchmark Carbon against Argon on your real workloads. Measure cost per finished task, not per token, and budget for the December price. For everyday Gemini users on free or AI Pro plans, Carbon is not available yet, and Google has not said when, or if, that will change, as The Verge pointed out.

ModelInput (per 1M tokens)Output (per 1M tokens)Note
Gemini 4 Carbon (intro)$2$10Until Nov. 30, 2026
Gemini 4 Carbon (standard)$4$20From Dec. 1, 2026
Gemini 4 Argon (intro)$2$10Same intro rate as Carbon
Claude Fable 5.2$8$40Launched Nov. 9, 2026