Alphabet (GOOGL) stock slipped 0.48% on Tuesday as Google opened Gemini 4 Carbon, its most powerful AI model, to paying developers and businesses. The release arrives with an introductory price that will double at the start of December.

Introductory pricing is $2 per million input tokens and $10 per million output tokens, and it ends after November 30. Starting December 1, those rates rise to $4 and $20 — meaning the same workload run after that date bills at twice the per-token cost. Google has not said how much revenue it expects Carbon to bring in. The company detailed the rollout in an official developer blog post.

Google CEO Sundar Pichai announced the launch on X:

Gemini 4 Carbon is rolling out today. It's built for sustained, long-horizon tasks, reasoning deeply through complex software engineering, cybersecurity, and scientific discovery. Coming globally via the Gemini API in @GoogleAIStudio and Gemini for business, and for Ultra…

— Sundar Pichai (@sundarpichai) November 10, 2026

In its announcement, Google positioned Carbon for sustained, long-horizon work, where the model reasons through complex software engineering, cybersecurity and scientific discovery problems over extended sessions.

Carbon matches the launch pricing of Gemini 4 Argon, which stays Google's pick for everyday coding and office work. Carbon is built for harder jobs with many steps, such as hunting bugs in complex software or conducting scientific research, where sustained reasoning over long horizons matters most.

Why Carbon Can Cost More Than the Price Tag Says

The catch is in how the model works. Carbon spends more time on tough problems and checks its own answers. That takes more tokens — the units AI systems process — so a complex request could end up costing more on Carbon even at the same per-token price as Argon. The December step-up therefore has its biggest effect on complex jobs, where the doubled rates apply to the largest token totals.

Google says Carbon ranked first on three AI coding and automation tests, and it finished those tasks at less than half the cost of the next-best rival, based on the company's reported results. Google also calls Carbon its strongest model for long-running agent work, a category that covers software engineering, full-stack web development and scientific reasoning.

Cybersecurity is a central part of the pitch. Wiz, Google's security partner, used Carbon to find previously unknown software flaws at two major software providers in under two hours, according to the company. Early users also turned up zero-day vulnerabilities in a short span.

Developers get another tool alongside the model. Google is testing a feature called Swarm in its Antigravity platform, which lets one instruction go out to several AI agents at once. One agent might write code while another checks it for mistakes.

Cloud Numbers and Wall Street's View

The launch lands amid strong momentum for Google's cloud business. Google Cloud revenue rose 103% last quarter to $30.7 billion, while operating profit more than tripled to $11.1 billion. The unit ended the quarter with $561 billion in contracted future revenue, although that figure includes services beyond AI models.

Analysts rate GOOGL a Strong Buy, based on 30 Buys and four Holds over the past three months. The average price target of $398.30 implies 38.7% upside.

According to Google, Carbon is available in Google AI Studio and Android Studio on the paid API tier, and in Gemini for business. Google AI Ultra subscribers can try the model in both AI Studio and Antigravity. The higher pricing takes effect on December 1, leaving a roughly three-week introductory window, from the November 10 launch to November 30, to run the model at the lower rates.

Source: Blockonomi