Google has opened Gemini 4 Carbon to paying developers and businesses. It is the company’s strongest AI model for long, hard jobs — coding, research and cybersecurity work that runs for hours. The catch is the price: $2 per million input tokens and $10 per million output tokens until 30 November, then both prices double from 1 December.
Updated: 12 November 2026, 10:00 IST
Quick Read
- Gemini 4 Carbon opened to paying users on 10 November 2026, after its announcement on 28 October. Google calls it its strongest model for cybersecurity and long-horizon agent work.
- Intro price: $2 per million input tokens, $10 per million output tokens, until 30 November. From 1 December: $4 and $20. Cached input is $0.10 now, $0.20 from December.
- It is NOT on the free tier and NOT in AI Pro. Paying API users, Gemini for Business customers and AI Ultra subscribers get it. No hands-on side-by-side test against Claude Fable 5.2 exists yet.
What Carbon is for
Carbon is built for jobs that run long. Google DeepMind says it handles complex coding and debugging, financial research, scientific research and cybersecurity defence — tasks that need extended reasoning and repeated checks. It sits above Gemini 4 Argon in Google’s line-up, which Google says “remains suitable for most software engineering and knowledge work”.
Google announced Carbon on 28 October, in a post by Koray Kavukcuoglu, its senior vice president and chief AI architect. For two weeks only cybersecurity defenders could use it, through Google’s Fairwind Program. On 10 November Google opened it to everyone who pays.
The model takes in up to 1,048,576 tokens of context — roughly a whole codebase — and can write up to 65,536 tokens in one answer. It has three thinking levels: low, medium and high. Thinking tokens are billed at the output rate.
Who gets it — and who does not
From 10 November, Carbon is live on the Gemini API in Google AI Studio and in Gemini for Business, for paying developers and enterprises. It is also rolling out to all Google AI Ultra subscribers in the Gemini app, AI Mode in Search and Google Antigravity. GitHub Copilot paid plans get it too.
Two groups do not get it. Carbon is not on the free tier, and it is not in AI Pro — the $19.99-a-month US plan gets Argon, not Carbon. AI Ultra in the US starts at $99.99 a month; no India price in rupees has been published.
Also read: Claude Fable 5.2 out: 20% cheaper, 30% faster
The price, in plain numbers
One million input tokens is roughly 750,000 English words. At the intro price that costs $2, about ₹187, per Gadgets360. One million output tokens costs $10, about ₹936. Cached input — text you have sent once before — costs $0.10 per million tokens.
From 1 December the deal ends: input goes to $4 per million tokens and output to $20, cached input to $0.20. The same work costs exactly double. Google’s pricing page confirms 30 November as the end of the intro period.
For an Indian developer on the paid API, the line is simple: work you run in November costs half of what the same work costs in December. A big job run before 30 November saves real money.
Carbon vs Fable 5.2 — the honest comparison
Anthropic’s Claude Fable 5.2, which we covered yesterday, costs $8 per million input tokens and $40 per million output tokens. Carbon at its intro price of $2 and $10 is about four times cheaper per token. From 1 December, at $4 and $20, it is still half Fable’s price.
But price is not the whole story. On the Artificial Analysis Intelligence Index, Carbon scored 60 against Fable 5.2’s 65, and Carbon ranks fourth, not first. Google says Carbon ranks first on three other benchmarks — CursorBench 4.0, Perplexity’s WANDR and Zapier’s AutomationBench. Those are Google’s claims.
No hands-on side-by-side test of Carbon against Fable 5.2 exists yet. The price comparison above uses list prices only. Which model is better for your code is something only your own test can answer.
Also read: Meta Muse in India at Rs 1,799: what you get
What it means for Indian developers
ET and Gadgets360 both covered this on 11 November — it is a live India tech story, not a US-only release. Indian developers on the paid API get the $2/$10 rate only till 30 November; after that, the same work costs double.
The smart move is boring: test Carbon now, while it is cheap, on a real job you already have. If it beats Argon or Fable, lock in the work before December. If it does not, you learned that for $2, not $4. If you are on the free tier or AI Pro, nothing changes.
Also read: GTA 6 preload live: 212GB download before 19 Nov
FAQ
How much does Gemini 4 Carbon cost?
Until 30 November, $2 per million input tokens and $10 per million output tokens. From 1 December, $4 and $20. Cached input is $0.10 now, $0.20 from December. Thinking tokens are billed at the output rate.
Can I use Gemini 4 Carbon for free?
No. Carbon is not on the free tier of the Gemini API, and it is not in AI Pro. It is open to paying API developers, Gemini for Business customers, Google AI Ultra subscribers and GitHub Copilot paid plans.
Is Carbon better than Claude Fable 5.2?
On price, yes — about four times cheaper per token at its intro price, half the price from 1 December. On the Artificial Analysis Intelligence Index, Carbon scored 60 against Fable 5.2’s 65. No hands-on test exists yet: cheaper, but not proven better.
Desk-original added context
Our own price table, per million tokens: Argon — $2 input / $10 output; Carbon — $2 / $10 until 30 November, then $4 / $20 from 1 December; Fable 5.2 — $8 / $40. Our own worked example of what $10 buys: at the intro rate, $10 buys 5 million input tokens or 1 million output tokens; from 1 December, the same $10 buys 2.5 million input tokens or 500,000 output tokens — exactly half. A team that spends $100 a month on Carbon in November pays $200 in December for the same work. Our own timeline: 28 October 2026 — Google announces Carbon; 10 November — Carbon opens to paying users; 30 November — intro price ends; 1 December — price doubles to $4 / $20.
Sources
- Google DeepMind blog — Gemini 4 Carbon is now available (10 Nov 2026, 200 verified; primary source — announcement, availability, intro price)
- Google AI Studio — Gemini API pricing (200 verified; intro price through 30 Nov, $4/$20 from 1 Dec)
- Economic Times — Gemini 4 Carbon opens to developers, enterprises with advanced AI capabilities for complex tasks (11 Nov 2026, 200 verified)
- Gadgets360 — Gemini 4 Carbon now available to developers and Google AI Ultra subscribers (11 Nov 2026; page loads — verified via browser, curl returned 403 bot-block; Rs 187/936 conversions)
Sasanka Sekhar Pradhan is the founder of India Presents, an India-focused news site. He covers daily box-office trackers, India news, economy explainers, sports previews and festival guides — in simple English, so readers across India can follow fast-moving stories without jargon. His focus is what a story means for the reader: the number that matters, the source behind it, and what happens next. India Presents publishes daily, with day-trackers updated as new figures come in.
