Google Rolls Out Gemini 4 Carbon, Its Strongest AI Model, at a Quarter of the Price of Claude Fable 5.2

AI
Written by Methodical Millions
Colourful editorial illustration of a glowing translucent shield around a crystalline hexagonal core, with blue, red, yellow and green data streams and neural network nodes on a dark navy background

Google's newest frontier model is now open to paying developers and AI Ultra subscribers. The bigger story is the price: an introductory $2 per million input tokens, about a quarter of what Anthropic charges for its newest model.

By Methodical Jon | Nov 10, 2026

Google (NASDAQ: GOOGL) made Gemini 4 Carbon widely available on Tuesday, November 10th 2026, two weeks after first releasing it to a small group of cybersecurity defenders. Google calls Carbon its "strongest model," with frontier performance in cybersecurity and long-horizon agent work across coding, knowledge work and science, according to the company's announcement. It costs $2 USD per million input tokens and $10 USD per million output tokens through November 30, the same introductory price as Gemini 4 Argon.

What happened

Carbon is now in the Gemini API through Google AI Studio and Gemini for business for all paying developers and enterprises. It's also rolling out to Google AI Ultra subscribers in the Gemini app, AI Mode in Search and Antigravity, Google's agent-first coding platform, The Verge reported. The Verge noted it isn't clear if Carbon will reach cheaper plans.

Ultra isn't cheap: about $100 USD a month to start, per Engadget and VentureBeat.

Google isn't replacing Argon, the Gemini 4 model it launched last month. The company says Argon "remains the ideal choice for the majority of software engineering and knowledge work," while Carbon is built for the hardest jobs, like multi-step coding and debugging and open-ended scientific research. Carbon uses more tokens than Argon to reason and check its own work. In the Gemini app, Carbon is labelled "Most complex work," per 9to5Google.

The price war, in numbers

The Verge says Carbon's introductory price is around a fifth of the cost of OpenAI's GPT-6 Astra and a quarter of Anthropic's Claude Fable 5.2, which launched a day earlier. VentureBeat makes the same quarter-of-Fable comparison.

The discount has an end date. After November 30, the price doubles to $4 USD per million input tokens and $20 USD per million output tokens, Google says. VentureBeat notes that's still half of Fable 5.2's standard rates.

There's a catch. The Verge points out that a lower per-token price doesn't always mean a cheaper job, because models use different numbers of tokens for the same task, and Google says Carbon uses more tokens than Argon. The Verge also warns that companies can cherry-pick benchmarks, and says the results Google cited at Carbon's October debut put it behind Fable 5.2 on most measures.

Meta also released Muse Ember, a 10-cent agent model, on Monday. That's three major launches from three labs in two days.

What Carbon is built to do

  • Agent swarms. Google is previewing a /swarm feature in Antigravity that turns a single prompt "into a coordinated team of agents," according to both Google and The Verge. Carbon takes the hardest roles, like architecture and auditing, while faster models handle the rest.
  • Benchmarks. Google says 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. These are Google's numbers.
  • Cybersecurity. Google says security firm Wiz used Carbon to find zero-day vulnerabilities at two major SaaS providers in under two hours. Engadget also reported that early users found zero-days quickly. Defenders can still apply for a cyber-specific version through Google's Fairwind Program.
  • Science. Google is expanding its program offering 10,000 scientists six months of free access to include Carbon. Engadget reports that thousands have already been onboarded.

Why it matters

For months, frontier AI looked like a two-horse race between Anthropic and OpenAI, The Verge wrote. Carbon puts Google back near the front, and it's using price to win share.

Cheaper top-tier intelligence changes the math for every company building on these models. A startup that couldn't afford a frontier model on every request might now. As per-token prices fall, the advantage moves to whoever owns the customer and the workflow.

The money lesson

1. Read the fine print on intro pricing. Carbon's price doubles after 20 days. Budget for the real price, not the launch price.

2. Measure cost per job, not cost per unit. A cheaper model that burns twice the tokens isn't cheaper. Compare the total cost to get the work done.

3. Don't marry one vendor. Three labs shipped major models in two days. Build so you can switch models when price or performance changes.

4. Small teams get bigger leverage. Agent swarms aim to give one founder the output of a whole team.

For a builder's view of shipping software on a small team, listen to Episode 97 - Spotlight: Yaro Bagriy - Founder of Evercodelab.com. Yaro talked about building a portfolio of small digital products mostly on his own, which is exactly the kind of builder that cheaper coding agents are aimed at.

What to watch

  • December 1 pricing. Will developers stay with Carbon once the price doubles?
  • Wider access. Google hasn't said if Carbon will reach AI Pro or free users.
  • Independent tests. How does Carbon compare on cost per task against Fable 5.2 and GPT-6 Astra?
  • The response. Will Anthropic and OpenAI cut prices to match?

At the time of writing (about 5:00 PM ET on Nov 10, 2026), Carbon is still rolling out, and Google hasn't published a country-by-country schedule, per VentureBeat.

Sources:

Thanks for reading another Methodical Minute. To read more just like this, please subscribe below and share with someone you know.

This article may be AI-assisted and is subject to errors and omissions.

Methodical Millions https://methodicalmillions.com
Next

Bitcoin Slips Below $73,000, Its Weakest Level Since August, as Hot CPI Puts a December Fed Hike on the Table