.png){kind=logo}
Google has started rolling out its latest artificial intelligence model, Gemini 4 Carbon, to the Gemini app, bringing more advanced reasoning and coding capabilities to subscribers of its AI Ultra plan. The new model is designed for complex tasks, including multi-step coding, debugging, and scientific research.
The rollout follows Google's preview of Gemini 4 Carbon in late October. The company is positioning the model as its most advanced option for tasks that require deeper reasoning and extended problem-solving.
Google AI Ultra subscribers can now access Gemini 4 Carbon through the Gemini app, AI Mode in Search, and Google Antigravity. The existing Gemini 4 Argon model will continue to remain available alongside Carbon.
Google says Argon remains the preferred choice for most software engineering and knowledge-based tasks, while Carbon is intended for more demanding work that tests the limits of AI reasoning.
In the Gemini app's model selection menu, Argon is listed under "Code & analysis", while Carbon appears under "Most complex work". With the latest rollout, Google has also removed Gemini 3.1 Pro from the app.
Gemini 4 Carbon is designed to handle complicated coding projects, software debugging and open-ended scientific research. Google says the model can also assist with legal, financial, and enterprise tasks by analyzing large volumes of information and generating spreadsheets, presentations, and reports.
The model is also designed to work with complex files containing charts, images, blueprints, diagrams and tables.
Google claims Carbon ranked number one in its new Multimodal Knowledge Agents (MKA) benchmark, developed with Artificial Analysis. The model used three times fewer tokens than the next-best model across 200 tasks assessing multimodal performance in agent-based workflows.
For software development, Carbon can run build-and-verify processes, identify errors, and examine both code and rendered user interfaces to produce functional applications.
Also Read: Google Photos Backup Arrives in Samsung Gallery With One UI 9.0
Alongside Carbon, Google is previewing a new /swarm feature in its Antigravity development platform. It allows developers to turn a single prompt into a coordinated team of AI agents that can be monitored in real time.
Developers can assign Carbon different responsibilities, including architectural planning, coordination, and auditing, while using faster models for parallel tasks.
Google has introduced Carbon at USD 2 per million input tokens and USD 10 per million output tokens until November 30. After the introductory period, pricing will increase to USD 4 per million input tokens and USD 20 per million output tokens.
The model is also available through the Gemini API paid tier in AI Studio and Android Studio, as well as Gemini for business and Google Antigravity.
Google is offering eligible scientists and researchers access to Argon and Carbon through Antigravity's highest usage tier free for six months, subject to eligibility, capacity, and selection.