Google Rolls Out Gemini 4 Carbon Globally for Complex AI Tasks
Google has made its Gemini 4 Carbon AI model widely available, targeting complex coding and scientific research. Initially released to select cybersecurity firms, it now serves all Google AI Ultra subscribers and enterprise developers. While Argon remains preferred for standard tasks, Carbon handles frontier intelligence challenges, with a new 'swarm' agent feature previewed on the Antigravity platform.
Key points
- Google, the tech giant, has globally released Gemini 4 Carbon, its most advanced AI model, following an initial limited launch to cybersecurity defense companies in late October.
- The model is designed for complex, multi-step coding, debugging, and open-ended scientific research, such as protein structure prediction, distinguishing it from the Argon model which remains ideal for standard software engineering.
- Access is now granted to all Google AI Ultra subscribers at $100 per month via the Gemini app, AI Mode in Search, and the Antigravity platform, while introductory API pricing starts at $2 per million tokens.
- Google is previewing a 'swarm' feature on Antigravity that allows users to turn a single prompt into a coordinated team of AI agents, assigning Carbon to roles requiring deep reasoning and auditing.
- The company has provided free six-month access to 10,000 scientists to aid research, noting that Carbon has already assisted in uncovering zero-day vulnerabilities and advancing medical discovery.
- While Carbon targets frontier intelligence limits, Google advises that Argon remains the better choice for the majority of knowledge work, and the older 3.1 Pro model has been removed from the Gemini app.
Global Availability of Frontier AI Model
Google has officially expanded the availability of its Gemini 4 Carbon model to a global audience, marking a significant shift from its initial restricted rollout. The model was first introduced in late October to a select group of cybersecurity defense companies, allowing early testing of its capabilities in identifying zero-day vulnerabilities. Now, the company is making the technology accessible to a broader range of users, including all Google AI Ultra subscribers and enterprise developers through Google AI Studio and Gemini for Business. This move positions Google to compete more directly with leading AI systems from competitors like Anthropic and OpenAI, claiming that Carbon pushes the frontier of intelligence further than previous iterations.
Distinction Between Carbon and Argon Models
Google has clarified the specific roles of its new models to guide user adoption. While Gemini 4 Carbon is designed for tasks that test the limits of a model’s intelligence, the company states that Gemini 4 Argon remains the ideal choice for the majority of software engineering and knowledge work. Argon, which launched last month, is optimized for standard coding and analysis tasks. In contrast, Carbon is built for complex, multi-step coding, debugging, and open-ended scientific research. In the Gemini app interface, Argon is categorized under "Code & analysis," while Carbon is labeled for "Most complex work." Notably, with this launch, Google has removed the older 3.1 Pro model from the Gemini app, signaling a consolidation around the newer Gemini 4 architecture.
Enterprise and Scientific Applications
The release includes specific provisions for scientific and enterprise use cases. Google is providing free access to Carbon for six months to 10,000 scientists, an initiative intended to support research in areas such as protein structure prediction, which is critical for medication development. The company notes that this access will presumably help train its models further. For legal, finance, and enterprise sectors, Carbon is touted for its ability to handle higher-precision knowledge work, including artifact generation and data analysis. The model is designed to follow strict schemas and formatting to generate polished spreadsheets, presentations, and reports, addressing the need for accuracy in professional environments.
Introduction of Swarm Agent Feature
To demonstrate Carbon’s utility in long-horizon work, Google is previewing a "swarm" feature on its agentic software development platform, Antigravity. This feature allows users to transform a single prompt into a coordinated team of AI agents that can be tracked in real time. Developers can assign Carbon to roles that demand the deepest reasoning, such as architectural design, coordination, and auditing. These roles are paired with faster, lighter models across parallel isolated workspaces and adversarial verifiers that test every change. This architecture aims to enhance efficiency and accuracy in complex development workflows by leveraging the specific strengths of different models within a unified system.
Pricing and Access Tiers
Google has maintained competitive pricing for the new model. The introductory price for using Carbon via the Gemini API is set at $2 per million tokens. Access is currently available to all paying developers and enterprises through Google AI Studio and Gemini for Business. For consumer users, access is restricted to Google AI Ultra subscribers, who pay $100 per month. These subscribers can use Carbon within the Gemini app, the AI Mode in Search, and on Antigravity. It remains unclear whether the model will be available to lower-tier subscribers or free users in the near future, maintaining a premium positioning for the most advanced capabilities.
Key takeaways
- Google has globally released Gemini 4 Carbon, moving from a limited cybersecurity pilot to broad availability for AI Ultra subscribers and enterprises.
- The model targets complex coding and scientific research, while Argon remains the recommended choice for standard software engineering tasks.
- A new "swarm" feature on Antigravity allows users to coordinate multiple AI agents, assigning Carbon to high-level reasoning roles.
- Google is offering free six-month access to 10,000 scientists to support research in areas like protein structure prediction.
- Access is restricted to Google AI Ultra subscribers at $100/month for consumer apps, with API pricing starting at $2 per million tokens.
- The older 3.1 Pro model has been removed from the Gemini app as Google consolidates its AI offerings around the Gemini 4 series.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.