Gemini 4 Carbon API Is Live: GA on 2026-11-10
As of 2026-11-12 you can call Gemini 4 Carbon: Google marked the model generally available (GA) on 2026-11-10, and both the official changelog and the official blog post say it opened to developers that day. The model ID is gemini-4-carbon, and the Gemini API models page now lists it next to gemini-4-argon; introductory pricing is $2 per 1M input tokens and $10 per 1M output tokens through 2026-11-30, then $4 / $20 from 2026-12-01.
Updated 2026-11-13
Claude, GPT and Chinese models on one key, billed per token; check /models for callable models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Four key points
General availability (GA) date
Google's official blog post opened Gemini 4 Carbon to developers that day, and the changelog records it as generally available (GA), calling it its strongest model for cybersecurity and long-horizon work.
The model ID in the Gemini API
The official model page gives the ID as gemini-4-carbon: a 1M context, 65,536 output tokens by default in synchronous calls, and up to 1,048,576 tokens with long decode.
Argon, at the same introductory price
The same section of the Gemini API models page still lists gemini-4-argon; Google describes Argon as the faster Gemini 4 frontier model with lower token spend per turn, while Carbon spends more thinking tokens on hard problems. Both carry the same introductory price through 2026-11-30. A third model, Gemini 4 Hydrogen (gemini-4-hydrogen), went GA on 2026-11-12 with its own pricing: a Free tier, and $1.50 / $7.50 per 1M input / output tokens on the paid tier.
Introductory price (per 1M input / output tokens)
Google says Carbon launches at the same introductory price as Argon: $2 input and $10 output per 1M tokens through 2026-11-30, then $4 / $20 from 2026-12-01; context caching is $0.10, $0.20 from 2026-12-01, implicit caching only.
Where the Gemini 4 Carbon API stands
As of 2026-11-12, Google marked Gemini 4 Carbon generally available (GA) in the official changelog and blog post of 2026-11-10 and opened it to developers, so you can call it on the paid tier of the API. In the Gemini API you can call Gemini 4 Carbon (gemini-4-carbon), the faster Gemini 4 Argon (gemini-4-argon) and, since 2026-11-12, Gemini 4 Hydrogen (gemini-4-hydrogen); Google positions Carbon as its strongest reasoning model, with a 1M context and thinking levels low / medium / high, defaulting to medium.
What Google's launch post says
On 2026-11-02, Google's October AI recap said Carbon was then in pre-release testing with the U.S. government and a subset of trusted cyber defenders through the Fairwind Program, while Google refined safety guardrails and expanded partner testing ahead of a broader launch to developers, businesses and consumers, and the recap gave no date. The official blog post fetched on 2026-11-12 (page dated 2026-11-10) is headlined “Gemini 4 Carbon is now available for developers.”: developers can build with it from that day, starting on the paid tier of the API in Google AI Studio and Android Studio, and Google AI Ultra subscribers can try it in AI Studio and Google Antigravity.
Official timeline
Gemini 4 Argon (gemini-4-argon) became generally available (GA) in the Gemini API, open to developers and enterprises.
Google announced Gemini 4 Carbon, available from that week to a subset of cyber defenders, with more trusted partners phased in through the Fairwind Program.
Google marked Gemini 4 Carbon (gemini-4-carbon) generally available (GA) and opened it to developers, with the same-day blog post explaining how to start on the paid tier of the API.
Officially confirmed vs limits to know
Officially confirmed
All of this comes from Google pages fetched on 2026-11-12: (1) the official changelog lists Gemini 4 Carbon (gemini-4-carbon) as generally available (GA) on 2026-11-10; (2) the official blog post of the same date says developers can start on the paid tier of the API in Google AI Studio and Android Studio; (3) the model page gives a 1M context, 65,536 output tokens by default, long decode up to 1M, and thinking levels low / medium / high; (4) the pricing page gives $2 input and $10 output per 1M tokens through 2026-11-30, then $4 / $20 from 2026-12-01; (5) it runs on the paid tier only, and every token is billed, including thinking tokens.
⚠️ Official limits to know
(1) The developer API opening date: Google has now given it, 2026-11-10, so no guesswork is needed. (2) Carbon's model ID, context length and limits in the Gemini API: all published now, with the ID gemini-4-carbon, a 1M context, 65,536 output tokens by default and up to 1,048,576 with long decode. (3) When Carbon's introductory price ends: the pricing page gives 2026-11-30, with new rates from 2026-12-01. (4) Paid tier only: the blog post says to start on the paid tier of the API, and prices are given on the paid tier. (5) No explicit caching: the migration notes say cachedContents / caches.create are not supported; implicit caching is on by default for prompts of at least 4,096 tokens, and cached input tokens are billed at 5%. (6) Capabilities it does not support: the model page marks image generation, audio generation and the Live API as not supported. (7) Output budget: Google recommends at least 32,768 for max_output_tokens.
Argon vs Carbon: both are open to developers now
Gemini 4 Argon (open)
Generally available in the Gemini API since 2026-10-13 as gemini-4-argon; the official model page describes it as the faster Gemini 4 frontier reasoning model with lower token spend per turn; its introductory price runs through 2026-11-30, then $4 / $20 from 2026-12-01.
Gemini 4 Carbon (open to developers)
Announced on 2026-10-28 and generally available (GA) on 2026-11-10; Google calls it its strongest model for cybersecurity defense and long-horizon engineering, its ID is gemini-4-carbon, and it spends more thinking tokens than Argon on hard problems.
What to do now
(1) To see how Carbon is called: the Gemini API models page, pricing page and changelog all list it now, the model ID is gemini-4-carbon, and it runs on the paid tier; the official example sets model to gemini-4-carbon and calls interactions under generativelanguage.googleapis.com. (2) To use Gemini 4 today, use gemini-4-carbon as the official catalog lists it; switch to gemini-4-argon when you want faster responses and lower token spend per turn, with the same call pattern. (3) Before you integrate, change two things per the official migration notes: set max_output_tokens to at least 32,768 and remove calls that create or reference explicit caches (cachedContents / caches.create); price your work by date, introductory pricing through 2026-11-30 and $4 input / $20 output per 1M tokens from 2026-12-01.
On QCode
For similar coding and agent work, you can use the Claude or GPT models on sale on QCode’s /models, such as claude-opus-5-5 and gpt-6.1-sol, billed per token with unit prices on /models. Getting started in three steps: (1) create an account; (2) top up (Alipay, WeChat Pay, cards, crypto); (3) for Claude models in Claude Code, set ANTHROPIC_BASE_URL to https://api.qcode.cc/api. Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days; if you already have a Personal Plan, add a booster pack when you need more quota.
Frequently asked questions
Is Gemini 4 Carbon out?
Yes. As of 2026-11-12, Google marked Gemini 4 Carbon (gemini-4-carbon) generally available (GA) on 2026-11-10, the changelog now carries that model ID, and the same-day blog post opened it to developers, so you can call it on the paid tier.
When will Gemini 4 Carbon be available to developers?
Google has now given the date: the Gemini API changelog records 2026-11-10 as the GA date, and developers can call gemini-4-carbon on the paid tier from that day. The 2026-11-02 October recap carried no date, and that stage is over.
How is Gemini 4 Carbon different from Argon?
Google positions them differently: Argon became generally available to developers on 2026-10-13, and Google describes it as faster with lower token spend per turn; Carbon became generally available on 2026-11-10, and Google calls it its strongest for cybersecurity defense and long-horizon engineering, spending more thinking tokens and more latency on hard problems. Both carry the same introductory price.
How much will the Gemini 4 Carbon API cost?
Google now publishes the paid-tier prices: $2 input and $10 output per 1M tokens through 2026-11-30; after the introductory period it is $4 / $20, from 2026-12-01. Context caching is $0.10 per 1M tokens, $0.20 from 2026-12-01, implicit caching only; Batch and Flex are $1 input and $5 output, then $2 / $10 from 2026-12-01.
Can I use Gemini 4 Carbon through QCode?
No. As of 2026-11-12, QCode does not offer the Gemini series, so no Gemini 4 model (Carbon, Argon or Hydrogen) is on /models; for similar work, use the Claude or GPT models on sale on /models.
How do I get Gemini 4 Carbon developer access?
There is no stand-alone application route for general developers: the official blog post says developers can build with Gemini 4 Carbon from 2026-11-10, calling gemini-4-carbon on the paid tier of the API directly, or starting in Google AI Studio and Android Studio; Google AI Ultra subscribers can also try it in AI Studio and Google Antigravity. Follow Google's official announcements for the entry point.
Sources
Google's official blog post “Gemini 4 Carbon is now available for developers.” (page dated 2026-11-10), Google's October AI recap (page dated 2026-11-02), the GA entry of 2026-11-10 in the Gemini API changelog, the Gemini 4 Carbon model page, the 2026-11-12 Gemini 4 Hydrogen entry in the changelog, the Gemini API pricing page and the official migration notes were all fetched and archived on 2026-11-12; the QCode side comes from qcode.cc /models, the pay-as-you-go page and the docs.qcode.cc access page (2026-11-12).
Use the Claude and GPT models on sale today
One QCode key calls the Claude and GPT models on sale on /models, billed per token with unit prices on /models.
Related reading
Pay-as-you-go Packs
Buy a one-time quota instead of a monthly plan and use it within 7 days; per-model rates are on /models.
Gemini 4 Pro Status Tracker
The official docs have no model called Gemini 4 Pro; the models listed under Gemini 4 on the official models page are Carbon, Argon and Hydrogen.
Gemini 4 Hydrogen GA: $1.50 / $7.50
GA on 2026-11-12, model ID gemini-4-hydrogen; official price $1.50 / $7.50, a Free Tier and a 1,048,576 unified context; how it differs from Carbon and Argon and how to call it.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
Checked on 2026-11-12 against the Google blog post, October recap, Gemini API changelog, model page, pricing page and migration notes fetched that day. All prices are Google's official prices and follow its official pages; QCode's available models and unit prices follow /models.