Google opens access to Gemini 4 Carbon: what the new AI model can do

Google has opened access to its new flagship model, Gemini 4 Carbon. The company says it was built for complex tasks in programming, cybersecurity, data analysis and scientific research.

The model can perform multi-step tasks, verify its own outputs, and collaborate with other AI agents.

What Gemini 4 Carbon can do

Google describes Carbon as a model for tasks that require prolonged analysis and complex decision-making. Unlike Gemini 4 Argon, which is geared toward everyday programming and working with information, Carbon spends more computational resources on searching for solutions and validating results.

In particular, the model can work with large software projects, find bugs in code, fix them and check whether changes introduced new problems.

For web development, Carbon can create applications, run them, analyze the output and autonomously fix discovered errors. The model also works with 3D graphics and game development.

Separately, Google introduced the /swarm feature in the Antigravity development environment. It lets you launch multiple AI agents to jointly tackle a single task. For example, a developer can assign Carbon to design an app’s architecture and audit the code, while simpler parts of the work are delegated to faster models. Agents run in parallel, and the user can monitor their progress in real time.

Beyond programming, Carbon can analyze large volumes of documents, spreadsheets, charts and images. Google notes the model can prepare financial reports, presentations and spreadsheets according to specified requirements.

According to testing results published by the company, Gemini 4 Carbon ranked first in CursorBench 4.0, Perplexity WANDR and Zapier AutomationBench. Google claims that in these tests the cost to complete a single task was more than twice as low as the nearest competing model.

One of the main development directions for Gemini 4 Carbon — is cybersecurity. The model can analyze software, find vulnerabilities, check whether they can be exploited and suggest fixes.

For example, during testing as part of the Wiz Scan for Good program, Carbon found previously unknown zero-day vulnerabilities in two major cloud software vendors in less than two hours.

Also, the Google Cloud Vulnerability Research team used Carbon to confirm that two bugs in popular open-source software could be exploited for attacks. 

In an internal test of the Project Fortify system, the model achieved 100% recall and 95% precision when validating discovered issues. These metrics apply to Google’s specific test set, not to all possible vulnerabilities.

Google will provide Carbon’s specialized cybersecurity capabilities to vetted professionals through the Fairwind program.

Google also tested Gemini 4 Carbon in scientific research. In particular, researchers used the model to search for related protein sequences in large databases. In an experiment with seven protein families, Carbon on average found 40 times more such sequences. This could help other models predict protein structure more accurately.

Additionally, Carbon was used to analyze lab notebooks and experimental data. The model helped identify errors in lab procedures and improve research methods.

Where Gemini 4 Carbon is available and how much it costs

Developers can access Gemini 4 Carbon via a paid API in Google AI Studio and Android Studio, and companies can get it through Gemini for business.

For regular users, the model is gradually becoming available in the Gemini app and AI Mode in Google Search with a Google AI Ultra subscription. It is also appearing in Google Antigravity.

An introductory API price will be in effect until November 30, 2026:

• $2 per million input tokens;
• $10 per million output tokens.

After the introductory period ends, the price will increase to $4 per million input tokens and $20 per million output tokens.

Google notes that Carbon is primarily intended for the most complex tasks, while Gemini 4 Argon remains the main model for most everyday workflows.

Read also: Instagram will use AI to detect teenagers worldwide who hid their age: what will change