Quick Answer
Gemini 4 Carbon (gemini-4-carbon) is Google’s most powerful AI model, built for the hardest, longest jobs: complex multi-step coding and debugging, open-ended scientific research, and cybersecurity defence. It has a 1,048,576-token context window, three thinking levels, and costs $2 per million input tokens and $10 per million output tokens through November 30, 2026 — after that, the price doubles to $4 and $20 from December 1. It is available on the paid tier of the Gemini API in Google AI Studio, and is rolling out to Google AI Ultra subscribers in Antigravity, the Gemini app, and AI Mode in Search. It is not on the free tier.
What Is Gemini 4 Carbon?
Gemini 4 Carbon is Google’s strongest reasoning model, announced as the top of the Gemini 4 family. It builds on Gemini 4 Argon’s foundation with extra post-training for deep reasoning, extended verification, and multi-step agent workflows. Google positions Carbon for tasks that test the limits of a model’s intelligence — the long, complex, high-stakes work where a weaker model would give up, cut corners, or need constant supervision.
Carbon is not a replacement for Argon. Google’s own guidance is clear: Argon remains the right pick for most software engineering and knowledge work. Reach for Carbon when the job is long, complex, or high-stakes — and expect a bigger bill, because it thinks longer.
Gemini 4 Carbon Release Date and Availability
Gemini 4 Carbon was first previewed on October 28, 2026, in a limited rollout to cybersecurity defenders through Google’s Fairwind Program. Wider access opened on November 10, 2026, when Google made it generally available to all paying Gemini API developers and enterprises, and started rolling it out to Google AI Ultra subscribers.
On the API, Carbon is available on the paid tier in Google AI Studio, and through Vertex AI. For subscribers, it is rolling out to Google AI Ultra in Antigravity, the Gemini app, and AI Mode in Search. In the model picker, Argon is labelled “Code & analysis” and Carbon is labelled “Most complex work”. Carbon is not available on the Gemini API free tier, so every processed token is billed. Availability can vary by product and account settings.
Gemini 4 Carbon Features
1. Deepest reasoning in the Gemini family
Carbon builds on Gemini 4 Argon’s foundation with extra post-training for deep reasoning, extended verification, and multi-step agent workflows. Google calls it its strongest model, and positions it for tasks that test the limits of a model’s intelligence.
2. Huge 1M-token context window
A unified context of 1,048,576 tokens, shared by input, reasoning, and output. That is enough to hold very large codebases, long documents, and long research material in a single conversation.
3. Adjustable thinking levels
You can choose low, medium, or high thinking. Medium is the default. Higher thinking means deeper reasoning — but also more thinking tokens, which are billed as output, so it costs more and takes longer.
4. Long outputs
Responses stop at 65,536 output tokens by default, with long decode able to push a single response much further, up to around 1M tokens. That makes Carbon suitable for extended code generation, detailed reports, and long-form research output.
5. Multimodal input
Text, images, video, audio, and PDF go in; text comes out. It also supports function calling, structured outputs, code execution, computer use, file search, and Google Search grounding.
6. /swarm in Antigravity (preview)
Google is previewing /swarm in Antigravity, which turns one prompt into a coordinated team of agents: a coordinator assigns specialised workers in isolated Git worktrees, while adversarial verifiers test and challenge every change.
7. Stronger cybersecurity safeguards
In simulated cyberattack testing, Carbon’s safeguards blocked 99% of attack trajectories from an adaptive red-teaming agent, cutting the per-attempt breach rate to 0.8%, down from 3.1% on Argon.
Gemini 4 Carbon Pricing Explained
Google prices Gemini 4 Carbon by token usage. The following are Gemini API prices per one million tokens, not subscription fees. The $2 / $10 price is an introductory offer that ends on November 30, 2026.
| Billing item | Through Nov 30, 2026 | From Dec 1, 2026 |
|---|---|---|
| Input, per 1M tokens | $2 | $4 |
| Output, per 1M tokens | $10 | $20 |
| Cached input, per 1M tokens | $0.10 | $0.20 |
Important details: thinking tokens are billed as output tokens, so deep-reasoning runs cost more than the input price alone suggests. Carbon is not available on the Gemini API free tier — every processed token is billed. Google AI Ultra usage works on the subscription’s usage allowance, not this per-token price. Pricing and product availability may change, so check Google’s official documentation before making deployment decisions.
Gemini 4 Carbon vs Gemini 4 Argon
| Gemini 4 Carbon | Gemini 4 Argon | |
|---|---|---|
| Position | Strongest model, for the hardest work | Ideal choice for most coding and knowledge work |
| Best for | Complex multi-step coding and debugging, open-ended scientific research, cybersecurity defence | Interactive coding, routine feature work, fast production loops |
| Context window | 1,048,576 tokens | 1,048,576 tokens |
| Per-token price | $2 / $10 (introductory) | Same per-token price |
| Thinking tokens | Spends more thinking tokens per turn than Argon | Fewer thinking tokens per turn |
| Artificial Analysis Intelligence Index v5.0 | 59.4 | 50 |
| Availability | Paid API + AI Ultra (rolling out) | AI Pro and up |
For readers comparing assistants more broadly, see our Gemini vs ChatGPT guide and our ChatGPT vs Gemini vs Claude comparison. You can also read our What Is Gemini 4 Argon? explainer.
Gemini 4 Carbon Benchmarks
- Artificial Analysis Intelligence Index v5.0: 59.4 for Carbon, vs 50 for Argon.
- GraphWalks long-context test (256K–1M tokens): 96.2% for Carbon, vs 84.2% for Argon.
- Multimodal agentic workflows: Carbon ranks #1 while using 3x fewer tokens than the next best model across 200 tasks, according to Google.
- Cybersecurity defence: safeguards blocked 99% of attack trajectories in simulated testing; breach rate 0.8% vs 3.1% on Argon.
These are Google’s and benchmark publishers’ own figures, not independently reproduced results. They should not be interpreted as guaranteed performance on everyday tasks.
How to Use Gemini 4 Carbon
1. In Google AI Studio (paid tier)
Open AI Studio, pick the paid tier, and select the model ID gemini-4-carbon. Set the thinking level (low / medium / high) for the job.
2. Through the Gemini API
Use the model ID gemini-4-carbon in your API calls. It works in Android Studio and Gemini for business as well.
3. In the Gemini app / AI Mode / Antigravity (AI Ultra)
Ultra subscribers get Carbon in the Gemini app, AI Mode in Search, and Antigravity. In the model picker, Argon is labelled “Code & analysis” and Carbon is labelled “Most complex work”.
4. With /swarm (preview, Antigravity)
In Antigravity, /swarm lets you split a development job across parallel agent workers, with verifiers checking the work.
Who Should Use Gemini 4 Carbon?
- Developers working on large migrations, multi-package changes, cross-subsystem debugging, or complex architectural redesigns.
- Researchers doing open-ended scientific work with very long documents or datasets.
- Security teams who need strong reasoning plus strong safeguards.
- Enterprises doing high-precision knowledge work — spreadsheets, presentations, and reports that must follow strict schemas and formatting.
For everyday coding, quick features, and routine knowledge work, Argon is still Google’s recommendation — it is faster and cheaper in practice because it spends fewer thinking tokens.
Frequently Asked Questions
What is Gemini 4 Carbon?
Google’s strongest AI model, released November 10, 2026, for deep reasoning, agentic coding, scientific research, and cybersecurity defence.
When was Gemini 4 Carbon released?
November 10, 2026, for paying developers and enterprises, with a rollout to Google AI Ultra subscribers. It was previewed on October 28, 2026, for cybersecurity defenders only.
How much does Gemini 4 Carbon cost?
$2 per million input tokens and $10 per million output tokens through November 30, 2026, then $4 and $20 from December 1, 2026. Cached input is $0.10, then $0.20.
Is Gemini 4 Carbon free?
No. It is not on the Gemini API free tier. It is available to paying API developers and to Google AI Ultra subscribers.
What is the context window of Gemini 4 Carbon?
1,048,576 tokens (about 1M), unified across input, reasoning, and output.
What is the model ID for Gemini 4 Carbon?
gemini-4-carbon on the Gemini API and Vertex AI.
Gemini 4 Carbon vs Gemini 4 Argon — which is better?
Carbon is stronger for the hardest, longest jobs. Argon is better for most everyday coding and knowledge work, because it is faster and spends fewer thinking tokens. Both cost the same per token.
What is /swarm in Gemini 4 Carbon?
A preview feature in Antigravity that splits one prompt into a team of AI agents working in parallel, with verifiers testing each change.
Final Verdict
Gemini 4 Carbon is Google’s strongest model for the hardest work — long migrations, open-ended research, and high-stakes reasoning — with a 1M-token context, adjustable thinking, and a $2 / $10 introductory price that doubles on December 1, 2026. For most everyday coding and knowledge work, Argon remains the faster, cheaper-in-practice choice. If your job is long, complex, or high-stakes, Carbon is the model to reach for — just watch the calendar, because the introductory pricing ends on November 30.
Editorial update: November 11, 2026. This article is based on Google’s launch coverage and model documentation.