Although Gemini 4 Carbon is not included in Google AI Pro, here are five ways to use it.
Gemini 4 Carbon is not in Google AI Pro. Use Ultra, multi-model apps, paid API, AI Studio, or OpenRouter; know costs and when Argon suffices.
Key Takeaways
- Gemini 4 Carbon is not included in Google AI Pro; consumer access currently requires Google AI Ultra, priced at $99.99 per month or higher.
- There are five practical ways to use Carbon: upgrade to Ultra, use a multi-model app, pay for API access in Google AI Studio, use OpenRouter, or rely on Gemini 4 Argon when sufficient.
- Paid API access to Carbon costs per token, with lower promotional rates before November 30, 2026, but long reasoning tasks may generate high output costs.
- For many reading, extraction, and summarization tasks, Gemini 4 Argon may already be enough, potentially saving $80 per month compared with upgrading to Ultra.
Meta description: Gemini 4 Carbon is not in Google AI Pro. See five access routes, $99.99 Ultra pricing, API costs, OpenRouter options, and when Argon is enough.
Tags: Gemini 4 Carbon, Google AI Pro, Google AI Ultra, AI pricing, OpenRouter
Featured image: an abstract, friendly illustration of tiered AI access gates, with no real people, no logos, and no text.
For Google AI Pro subscribers, Google released Gemini 4 Carbon to the public on November 10, 2026, but the model is not included in the $19.99 plan. As of November 12, 2026, Carbon sits behind Google AI Ultra at $99.99 a month and up, while Google AI Plus includes neither Carbon nor Gemini 4 Argon. This report maps five ways to use Carbon now, what each costs, and when Argon is enough.
Google’s Varun Mohan wrote on X on November 11 that the team is still working out a "broader rollout." Until that happens, the consumer path in Google’s own apps remains Ultra, and no date has been given for Pro.
What Google AI Pro users see today
A Pro subscriber who taps Carbon in the Gemini app may be sent to an upgrade screen for $99.99 a month, not to the model. That is the current access boundary. In the Gemini app, AI Mode in Search, and Antigravity, Carbon is an Ultra feature. The AI Studio plans page states the same limit for developers: Carbon requires an AI Ultra plan in AI Studio and is not available on Free or AI Pro.
The paid Gemini API and Gemini for business do include Carbon, but the free tier does not. That leaves five practical routes for people who need it now.
Five ways to use Gemini 4 Carbon now
1. Upgrade to Google AI Ultra. The Ultra route costs $99.99 a month and up, $80 more than Pro. It unlocks Carbon across Google’s consumer and developer surfaces, including the Gemini app, AI Mode in Search, Antigravity, and AI Studio, with higher limits than third-party apps. If Gemini is the only model a person uses, this is the simplest answer.
2. Use a multi-model app that already includes it. Some paid multi-model apps list Gemini 4 Carbon, usually at a higher credit cost per message than Argon. HaloMate, for example, placed Carbon beside Argon and Gemini 3.1 Pro, as well as models from OpenAI, Anthropic, and other providers. In that setting, a user can regenerate the last answer with Carbon after working with Argon, compare the results side by side, and keep context while switching. The author’s HaloMate Max plan costs about the same as Google AI Ultra; if Gemini were the only model used, Ultra would provide more Carbon. The value is choosing across providers when models move or retire, as happened with GPT-4o and Opus 4.5. Before choosing any app, check whether Carbon is included in the plan or billed on top, and whether files and instructions carry over when switching models.
3. Use Carbon in Google AI Studio with a paid API key. Signing in to AI Studio with Pro does not make Carbon selectable, because subscription access follows the plan. Creating an API key and turning on Cloud Billing allows Carbon calls in the same interface, paid per request. The price is $2 per million input tokens and $10 per million output tokens until November 30, 2026, then $4 and $20. Carbon tends to think longer than Argon, and thinking tokens count as output, so long answers may cost more than they look. This spend is on top of $19.99 for Pro, and the result is a developer console rather than a workspace: sessions do not carry over unless the user builds that structure.
4. Use Gemini 4 Carbon through OpenRouter. OpenRouter added Carbon on launch day, served through Google AI Studio and Vertex at Google’s API rates. Users load credit in advance and pay per token. For someone already using OpenRouter with a chat client, it is a fast way to try Carbon for a few dollars with no monthly commitment. For someone without a client, it requires setup, bills separately from existing subscriptions, and does not provide shared project memory. It is better for quick experiments than for work that needs to stay organized over weeks.
5. Stay on Gemini 4 Argon and switch only when needed. This costs $0 extra on Google AI Pro. Google’s launch post says Argon remains the ideal choice for the majority of software engineering and knowledge work, while Carbon is built for tasks that test the limits of a model’s intelligence. Artificial Analysis scores match that split: Carbon leads on the overall Intelligence Index, 60 to 50, driven by hard coding and reasoning, but long-context document reasoning is nearly tied at 79.3% for Carbon and 79.0% for Argon.
In the author’s test, a 60-page Stanford report was loaded into a Project. Argon handled the reading and fact-checking, including accurate chart-bar extraction, and reached the same conclusion as Carbon. When both models were asked to turn the findings into a short animation for a meeting, Argon made a cramped slide deck and recorded its own play buttons into the video; Carbon made a version the author would play in the room. If the task is reading, extracting, and summarizing, Pro may already be enough. Running a hard task on Argon first can save $80 a month.
Other access paths and platform notes
Perplexity makes Carbon available inside Perplexity Computer on its Pro and Max plans, and in regular search on Max only. Cursor has had Carbon since launch day. GitHub Copilot includes Carbon, with Business and Enterprise admins controlling access. Google is also taking applications from scientists and researchers for six months of free access through Antigravity’s top tier, subject to selection.
The report notes that Jio-bundled Google AI Pro users may currently sit a tier below regular Pro, creating a separate access problem. It also compares Carbon with Claude Fable 5.2 on the API: VentureBeat’s price comparison lists Fable 5.2 at $8 per million input tokens and $40 per million output tokens, making Carbon a quarter of that price before November 30, 2026, and half after.
Why this matters beyond Carbon
The access pattern around November 10 was not only about a new model. As Carbon rolled out, the Gemini app started removing Gemini 3.1 Pro, a model some users had built prompts around. Anthropic is described as preparing a similar tier shift with Fable on Claude Pro on November 17.
The broader lesson is that plan boundaries can move while daily work remains tied to one vendor’s app. The strongest model can move up a tier, an older model can disappear, and the users most affected are those whose context lives only inside that provider. The report’s practical advice is to keep instructions, files, and context in one place and treat the model as the part that changes, because it is likely to change again before the end of the month.
The original report was published in Towards AI on Medium.

Comments (0)