Gemini 4 Carbon is generally available
TL;DR
On November 10, 2026 Google made Gemini 4 Carbon generally available as gemini-4-carbon. Introductory API pricing is $2 per million input tokens and $10 per million output tokens through November 30, then $4 and $20. Google AI Ultra subscribers are getting it in the Gemini app, AI Studio, and Google Antigravity.
- Status
- Live
- Released
- Plans
- Paid Gemini API; Google AI Ultra in the Gemini app, AI Studio, and Antigravity
What changed
Google's API release notes for November 10, 2026 mark gemini-4-carbon generally available. Google calls it its most capable reasoning model for sustained agentic coding, multimodal knowledge work, scientific reasoning, and cybersecurity defense.
The developer guide, published the same day, sets the price and the knobs:
- Introductory price through November 30, 2026: $2 per million input tokens and $10 per million output tokens. From December 1: $4 and $20. Cached input is 95% off. Google says this matches Gemini 4 Argon's introductory price.
- A 1 million token context window for input and output on the Gemini API.
thinking_leveloflow,medium, orhigh. Medium is the default. High is for deeper reasoning. Low is for latency and cost.- The same built-in Gemini API tools as the rest of the Gemini 4 family. Switching between Carbon and Argon is a model-string change.
Where it runs:
- Paid Gemini API in Google AI Studio, for paying developers.
- Google AI Ultra subscribers in AI Studio and Google Antigravity. The developer guide also says scientists and researchers can apply for six months of Antigravity access at no charge, subject to eligibility, capacity, and selection, and that cybersecurity defenders can apply to the Fairwind Program for Carbon for cyber.
- The Gemini app. @GeminiApp said on November 10 that Carbon is rolling out to Google AI Ultra subscribers on desktop, mobile, and the web at gemini.google.
Google first announced Carbon about two weeks earlier and released it to cyber defenders in the Fairwind Program. November 10 is the broader release. The developer guide says to keep using Gemini 4 Argon (gemini-4-argon) for most software engineering and knowledge work, and to use Carbon when the task needs longer exploration, heavier verification, or harder reasoning. In multi-agent setups it suggests Carbon for architecture, coordination, and final checks, and Argon or Gemini 3.8 Flash for parallel implementation.
Why it matters
The model string to ship against is gemini-4-carbon, and the price changes on December 1. Anything you budget at $2 / $10 has to be re-priced at $4 / $20 twelve days later, before cached-input discount. Argon stays the default for ordinary coding if you want the tighter token use Google describes.
Ultra subscribers see Carbon in the Gemini app as a rollout, not as an instant switch for every account. API access for paying developers is the part marked generally available in the changelog.
Fairwind is a separate, earlier door for cyber defenders. General availability does not by itself put Carbon for cyber into every project.
Frequently asked questions
What is the model ID and the price?
gemini-4-carbon. Through November 30, 2026, Google's developer guide says $2 per million input tokens and $10 per million output tokens. From December 1, $4 and $20. Cached input is 95% off.
How do I control how much it thinks?
Set thinking_level to low, medium, or high. Medium is the default.
Should every request move from Argon to Carbon?
Google's developer guide says Argon remains the model for most advanced software engineering and knowledge work. Use Carbon for the hardest multi-step, scientific, or security tasks.