Google makes Gemini 4 Carbon generally available with introductory pricing
Google says its Gemini 4 Carbon reasoning model is now generally available in the Gemini API at introductory prices through November 30, 2026.
Google has made Gemini 4 Carbon, model ID gemini-4-carbon, generally available in the Gemini API. Introductory prices are $2.00 per 1M input tokens and $10.00 per 1M output tokens through November 30, 2026, according to a Gemini API changelog entry dated November 10, 2026.
Google describes the model in that entry as its most capable reasoning model, built for sustained agentic coding, multimodal knowledge work, scientific reasoning and cybersecurity defense. Those are the company's own categories. The changelog offers no benchmarks or comparisons to support them, so the claim of capability is Google's rather than something the entry demonstrates.
The announcement gives the pricing, and the changelog labels it introductory. The rates run only until November 30, 2026. The entry does not say what the price will be after that date.
What changed
Gemini 4 Carbon (gemini-4-carbon) is now generally available in the Gemini API at introductory prices of $2.00 per 1M input tokens and $10.00 per 1M output tokens, with those rates running only through November 30, 2026.
Type: New capability · Applies to: gemini-4-carbon, Gemini API
Why it matters
Teams building agentic coding or multimodal tools on the Gemini API can now use the model in production. The introductory rates set a cost baseline that will change after November 30.
What you can do
Engineering leads should benchmark Gemini 4 Carbon against their current models and budget for the post-November 30 pricing before committing to it.
How the story developed
Google's own changelog gives the exact introductory prices of $2.00 per 1M input and $10.00 per 1M output tokens and lists the model's intended uses.
Theo's hands-on review adds reported false-claim rates, task latency, cache miss rates and the December 1 price increase.
Adds independent results: Vals ranks Carbon third, Roboflow ranks it last on speed and cost, and testers report hallucinations and overconfidence.
This post adds that the price rises to $4 input and $20 output per million tokens on December 1, and that the model was first announced two weeks earlier for a subset of cyber defenders.
Google DeepMind's own announcement confirms the availability, so the claim now has an official source.
Original source
Further reporting
- link.mail.beehiiv.comGoogle AI
- link.mail.beehiiv.comGoogle AI
- link.mail.beehiiv.comCopilot
- link.mail.beehiiv.comCursor
- link.mail.beehiiv.com@_philschmid
- link.mail.beehiiv.com@_philschmid
- link.mail.beehiiv.comCursor
- link.mail.beehiiv.comCognition
- link.mail.beehiiv.comPerplexity
- link.mail.beehiiv.comVals
- link.mail.beehiiv.comARC Prize
- link.mail.beehiiv.com@skalskip92
- link.mail.beehiiv.comArena
- link.mail.beehiiv.com@krishnanrohit
- link.mail.beehiiv.com@kimmonismus
- link.mail.beehiiv.com@OfficialLoganK
Where we saw it
Gemini API changelog, Google DeepMind, Latent Space, Simon Willison, Theo - t3.gg
AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works