Gemini 4 Carbon is now general: $2/$10 until Nov 30, then it doubles
Direct answer
Gemini 4 Carbon is now available to all paying developers as of November 10, 2026, according to Google's blog and the Gemini API changelog. The introductory price is $2 per million input tokens and $10 per million output tokens until November 30; from December 1 the price becomes $4 and $20. Cached reads cost $0.10 until November 30 and $0.20 from December. Carbon is the deeper model of the Gemini 4 family, built for coding, research and agent tasks, with context of up to 1 million tokens.
What changed on November 10
- From limited access to general: Google announced Carbon on October 28 and opened it first only to cybersecurity defenders in the Fairwind Program. On November 10 it opened the paid tier of the Gemini API, Google AI Studio and Android Studio, enterprises through Gemini for Business, and Google AI Ultra subscribers in the Gemini app, AI Mode in Search and the Antigravity agent platform.
- The same introductory price as Argon: Carbon launches at $2/$10 per million tokens, the same launch price as Gemini 4 Argon. The difference is that Carbon is the deeper, slower sibling: it uses more thinking tokens than Argon and offers configurable thinking levels to balance quality, latency and cost.
- The price doubles in three weeks: the introductory price ends November 30, 2026. From December 1, the global price becomes $4 input and $20 output per million tokens, with cached reads at $0.20. Any budget built on the November price is half the real December price.
- The Cyber Permissive variant already charges the December price: that Carbon variant is billed now at $4/$20, with no promotional price. That detail confirms the November price is a promotion, and the December price is the stable price of the model.
What Google claims about the model, and what is still unverified
- Google says it leads on three benchmarks: according to Google, Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench, at less than half the cost per task of the next best external model. These are Google's figures and its launch partners' figures; they are company claims, not independent measurements.
- Cost per task matters more than cost per token: The Verge notes that a model that thinks longer can cost more per task even at a lower price per token. Carbon spends more thinking tokens than Argon, and thinking tokens are billed as output. The honest comparison measures the cost of your real tasks, not the list price.
- Explicit caching is out: Carbon does not support explicit caching (cachedContents), and sampling settings are ignored, according to the documentation. Teams migrating from Gemini 3 or earlier models should read the migration notes before switching models in production.
- No free tier: Carbon is not on the API free tier. API access requires the paid tier, and access in the Gemini app requires the Google AI Ultra subscription.
What to do in your business this week
- Budget at the December price: calculate the cost of your tasks at $4 input and $20 output per million tokens. If the use case only works at the November price, the use case is fragile.
- Measure cost per task before migrating: run your real tasks on Carbon and on your current model, at the same thinking level, and compare the total cost per finished task. Price per token is catalog data; cost per task is your business data.
- Set the thinking level per task: Carbon lets you adjust thinking effort. Routine tasks run at low effort; research and agent tasks run at high effort. The thinking level is the lever that controls cost.
- Mark December 1 on your calendar: the price change arrives in two weeks. If your provider bills with Carbon, ask in writing how it passes the price change to your invoice.
Note: this article summarizes Google's announcement of November 10, 2026, the Gemini API changelog, and coverage by Promptea, The Verge, APIYI and Wokey. The benchmark and cost-per-task figures are claims by Google and its launch partners. Prices are Google's official list prices for the global tier.
Want AI that works for your business?
Message us on WhatsApp and let's talk about automation and AI agents for your business.
WhatsApp inicIARelated articles
- Google opened Gemini 4 Argon with launch pricing: $2 per million tokens until Nov 30
- Claude Haiku 5.5 costs up to 90% less: the AI price war continues (and benefits you)
- Google launched the universal Gemini agent: the digital coworker with its own email
Frequently asked questions
When did Gemini 4 Carbon become generally available?
On November 10, 2026. Google announced it on October 28 and opened it first only to cybersecurity defenders in the Fairwind Program; on November 10 it opened it to paying developers, enterprises and Google AI Ultra subscribers.
How much does Gemini 4 Carbon cost?
Until November 30, 2026, $2 per million input tokens and $10 per million output tokens, with cached reads at $0.10. From December 1, 2026, the price becomes $4 input, $20 output and $0.20 cached reads.
How is Carbon different from Argon?
Carbon is the deeper model of the Gemini 4 family: it uses more thinking tokens than Argon, offers configurable thinking levels and is built for coding, research and agent tasks. Argon is the general-purpose frontier model; Carbon is the deep-work model.
Does Carbon have a free tier?
The Gemini API does not offer Carbon on the free tier. API access requires the paid tier, and access in the Gemini app requires the Google AI Ultra subscription.
