
Gemini 4 Carbon is now available: who gets it and what's new
9 min read
Last updated November 11, 2026. How we cover news: we write from the official announcement and docs (linked below) and point you to the tools on Unloc. Not sponsored.

Short answer: Google made Gemini 4 Carbon generally available on November 10, 2026, two weeks after announcing it for vetted cyber defenders only. Developers and businesses can now use it on the paid Gemini API, in Google AI Studio and in Gemini for business at an introductory $2 per million input tokens and $10 per million output tokens until November 30. In the Gemini app, AI Mode in Search and Google Antigravity, it is rolling out to Google AI Ultra subscribers only. It is also in GitHub Copilot and Cursor.
Sponsored
Your tool here — sponsor this list
Fixed $149/month on Draft with AI. A subscription — not Claim #1.
Pin this category · $149/moUpdate, November 11, 2026: this page first covered Google's October 28 announcement, when almost nobody could use Carbon. Google has now opened it up, so we have rewritten the page around what changed and who can use it. It is based on Google's November 10 availability post, the Gemini 4 Carbon model card and the Antigravity and GitHub announcements, and we say where a figure is Google's own.
What Google announced
On November 10, 2026, Google said it was "making Gemini 4 Carbon available to developers, enterprises, scientists, and consumers". Google calls Carbon its strongest model, built for long-running agent work across coding, knowledge work, science and cybersecurity. It first went to a small group of security teams through Google DeepMind's Fairwind Program after the October 28 announcement.
The main points from the launch post:
- Carbon sits above Argon, not instead of it. Google says Gemini 4 Argon "remains the ideal choice for the majority of software engineering and knowledge work". Carbon is for the hardest jobs, such as complex multi-step coding and debugging or open-ended research, and it uses more tokens than Argon to explore, reason and check its own work.
- Third-party benchmark claims. Google says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench, at less than half the cost per task of the next best model from another company. It also says Carbon ranks second on Artificial Analysis's new Multimodal Knowledge Agent benchmark.
- Multi-agent coding in Antigravity. Google is previewing a
/swarmcommand in Google Antigravity that turns one prompt into a team of coordinated agents. - Stronger safety numbers, by Google's account. Google says Carbon's safeguards blocked 99% of attack attempts from its own red-teaming agent in simulated cyberattack tests and passed 92% of tasks on an internal agentic honesty test, up from 70% for Argon.
According to the model card, Carbon is built on Gemini 4 Argon. It takes text, images, audio and video, with a combined input and output context window of up to one million tokens.
Who can use Gemini 4 Carbon now
Access depends on how you use Google's AI. From Google's post and the model card:
- Developers: the paid tier of the Gemini API in Google AI Studio and Android Studio. The model ID is
gemini-4-carbon. Google AI Ultra subscribers can also try it in Antigravity and AI Studio. - Businesses: Gemini for business, through the API or the model picker, plus Google Antigravity and Google's CodeMender security agent.
- Everyday users: the Gemini app and AI Mode in Search, but only on Google AI Ultra. Google's subscriptions page lists Carbon under Ultra, and says Carbon in AI Mode is limited to select countries. Google AI Pro subscribers keep Argon as their top model.
- Scientists: Google is adding Carbon to its programme that gives selected researchers free access to Antigravity's highest usage tier for six months. Places depend on eligibility and capacity.
- Security teams: the Fairwind Program is still the route to "Carbon for cyber", the version whose security benchmark results Google quotes.
Outside Google, GitHub Copilot made Carbon generally available on the same day for Copilot Pro+, Max, Business and Enterprise users, in VS Code, JetBrains IDEs, the Copilot CLI, the coding agent and github.com, with a gradual rollout and usage billed at Google's list price. Cursor lists Gemini 4 Carbon on its models page as of November 11, 2026.
What it costs
On the API, Google is charging the same introductory price as Argon: $2 per million input tokens and $10 per million output tokens. A footnote in Google's post says that after the introductory period ends on November 30, 2026, the price becomes $4 per million input tokens and $20 per million output tokens. Check Google's Gemini API pricing page before you budget.
Keep in mind that Google says Carbon uses more tokens than Argon on the same task, so a job can cost more than the per-token price suggests. Google's cost-per-task claims come from the benchmarks it chose. For consumers, Carbon comes with Google AI Ultra, not with a separate fee; see Google's plans page for what Ultra costs where you live.
What's new in Antigravity: /swarm
The biggest product change is in Google Antigravity, Google's agent-first coding app. Typing /swarm hands your prompt to a coordinator agent that starts parallel worker agents in separate git worktrees, so their edits don't collide, plus verifier agents that run tests and challenge each change. A side panel shows the agents' shared message board, a roster you can use to step into any agent, and a timeline of how the team got to the final code.
You can pin each agent to a different model. Google suggests Carbon for the coordinator, architecture, security audit and verification roles, with lighter models doing routine edits. Google also warns that "parallel multi-agent systems consume quota much faster than a standard chat". Carbon is available to Ultra subscribers in the Antigravity app, CLI and IDE extensions, and /swarm is rolling out in preview in the app and CLI.
How Carbon compares with Claude and GPT
Carbon lands a day after Anthropic released Claude Fable 5.2, so the top of the market changed twice in two days. A few things matter when you are picking a model:
- Leaderboards have already moved. When we first wrote this page, Carbon led the Artificial Analysis Intelligence Index. On November 11, 2026, that index showed Claude Fable 5.2 ahead of it. Google's own launch claims focus on agent benchmarks run by Cursor, Perplexity and Zapier.
- Price per token. Carbon's introductory API price is well below Anthropic's listed $8 and $40 per million tokens for Fable 5.2. It doubles after November 30, and Google says Carbon uses more tokens than Argon, so compare cost per finished task, not just the rate card.
- Consumer access. Claude Fable 5.2 is on all paid Claude plans. In ChatGPT, OpenAI's GPT-6 family is the current line-up. Google keeps Carbon to its most expensive plan for now, so on Google AI Pro the fairer comparison is still Argon against Claude and ChatGPT.
- Coding tools. Carbon, Fable 5.2 and OpenAI's GPT-6 Astra all appear in GitHub Copilot's list of supported models, and Cursor's models page lists Carbon, Fable 5.2 and OpenAI models too. If you already use one of these tools, you can test Carbon against the others without switching apps.
All the benchmark numbers in Google's post are Google's own choice of tests, and many were run before release. Treat them as a starting point and test on your own work.
What it means if you're choosing AI tools
- Gemini app users: on Google AI Pro, nothing changes; Argon is still your best model. Upgrade to Ultra only if you hit Argon's limits on long, hard tasks, not just to get the newest name. See Claude vs Gemini and ChatGPT vs Gemini for the everyday differences between the apps.
- Developers on the Gemini API: try Carbon now while the introductory price lasts, but keep Argon as the default for routine calls and save Carbon for the steps where Argon falls short. Budget at the post-November price.
- Teams using coding agents: the quickest test is in a tool you already pay for. GitHub Copilot and Cursor both carry Carbon, so you can run it against Claude and GPT models on the same codebase. Our Cursor vs GitHub Copilot and Claude Code vs Cursor comparisons cover the tools themselves.
- Heavy agent users: Antigravity's
/swarmis the most distinctive new feature here, but it is an Ultra-only preview and burns through limits quickly. Teams who prefer a terminal agent can compare it with Claude Code.
The wider lesson is the same as in October: frontier models now leapfrog each other within days. Tools that let you switch models, such as GitHub Copilot and Cursor, protect you from betting on one provider. For what's next from Google's rivals, see our GPT-7 tracker.
FAQ
Is Gemini 4 Carbon available now?
Yes. Google made Gemini 4 Carbon generally available on November 10, 2026. Developers and businesses can use it on the paid Gemini API, in Google AI Studio and in Gemini for business, and it is rolling out to Google AI Ultra subscribers in the Gemini app, AI Mode in Search and Google Antigravity. It is also available in GitHub Copilot and Cursor.
Can I use Gemini 4 Carbon on Google AI Pro?
Not in the Gemini app for now. Google says Carbon in the Gemini app and AI Mode is for Google AI Ultra subscribers, and its subscriptions page lists Carbon under Ultra only. On AI Pro, Gemini 4 Argon remains the top model. Developers can still reach Carbon through the paid Gemini API without an Ultra plan.
How much does Gemini 4 Carbon cost?
On the Gemini API, Google charges an introductory $2 per million input tokens and $10 per million output tokens. Google says that after November 30, 2026, the price rises to $4 and $20. For consumers, it comes with the Google AI Ultra plan. Check Google's live pricing pages for current rates.
What's the difference between Gemini 4 Carbon and Gemini 4 Argon?
Carbon is built on Argon and aimed at the hardest tasks, such as long, multi-step coding, debugging and open-ended research. It spends more tokens reasoning and checking its work. Google says Argon is still the right choice for most software engineering and knowledge work, and Argon is available on more plans.
Is Gemini 4 Carbon better than Claude Fable 5.2 or GPT-6?
It depends on the task. Google says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench, but Artificial Analysis's overall intelligence index showed Claude Fable 5.2 ahead of it on November 11, 2026. Carbon is cheaper per token on the API for now. The best check is to run the same real tasks through each model, which GitHub Copilot and Cursor make easy.
Submit a tool if it belongs on the board. Listings are reviewed before they go live.
Share your learnings on socials
Other resources
View all
Best Calendly alternatives: free and paid scheduling tools
The best Calendly alternative depends on why you are leaving: Cal.com is the closest all-round swap with a generous free plan, SavvyCal gives guests a nicer booking experience, Google Calendar, Microsoft Bookings and HubSpot may already be in your stack, Chili Piper routes sales leads, and Acuity, Zoho Bookings and TidyCal handle paid client bookings.

Claude Fable 5.2 and Mythos 5.2: what's new and who it's for
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026: the same model, with Fable open to paid plans and the API and Mythos limited to verified organisations. Here's what changed, how it fits your Claude plan, what developers must update and how it compares with Opus 5.5, GPT-6 Astra and Gemini.

Best Ahrefs alternatives: free and paid SEO tools
The best Ahrefs alternative depends on the job: Semrush is the closest all-in-one swap, SE Ranking and Moz Pro are full suites many small teams find easier to run, Mangools and Ubersuggest are lighter keyword tools, and Majestic is a backlink specialist. For data on your own site, start with Google Search Console and Ahrefs Free.