Google has launched Gemini 4 Argon worldwide, expanding access to its latest AI model for developers, enterprises, and paying Gemini subscribers. Announced on October 13, 2026, the rollout brings Argon to Google’s developer and business platforms, with consumer access through the Gemini app and AI Mode in Search.

The release follows Google’s September 30 introduction, when access initially focused on trusted cybersecurity defenders through its Fairwind Program. Argon is designed to handle extended, multistep assignments across software engineering, business research, and security.

Built for Complex Development Work

Google’s developer announcement emphasizes code migrations and other engineering tasks that require sustained reasoning. Developers can access Argon through the paid Gemini API tier in Google AI Studio and Android Studio. Google AI Ultra subscribers can also use it in AI Studio and Google Antigravity, the company’s platform for development with AI agents.

Google reports that internal teams using Argon in Antigravity increased agentic code submissions by more than 35%, while code rollbacks declined by nearly one-third. These are Google’s internal results, rather than a guaranteed productivity improvement for other organizations.

Earlier examples included agents working on C/C++-to-Rust migrations. Google said critical rewrites undergo automated and manual auditing, emulation testing, and review before production deployment. That detail matters: generating a large code change and proving that it is safe to deploy remain separate engineering responsibilities.

Enterprise Access and API Pricing

Enterprise access includes Gemini for business, the API, and Google Antigravity. Google announced the following pricing for the public-access Argon model:

API usage

Introductory pricing through November 30, 2026

After the introductory period

Input tokens

$2 per million

$4 per million

Output tokens

$10 per million

$20 per million

Google also advertises a 95% discount on cached input tokens.

Beyond development, Google positions Argon for financial research, legal drafting, and work involving documents, charts, and video. Its earlier announcement highlighted evaluations covering software engineering and business task execution. These results describe performance on specific tests; they do not establish accuracy across every enterprise workflow.

Gemini App Adds More Task Capabilities

Google AI Pro and Ultra subscribers can now access Argon in the Gemini app. Google says the model provides more direct responses, reduces repetition, and communicates progress more effectively during longer assignments.

The app also introduces a switch between chat and tasks, allowing Gemini to continue working while users are away. Google says Gemini now connects with more than 40 apps, including Adobe, Shopify, Instacart, and OpenTable. Antigravity’s coding interface is also available directly within the Gemini macOS app.

Security Controls Address More Autonomous Agents

Google’s safety announcement describes protections that screen incoming requests, monitor signals inside the model, and check its activity as work progresses. The company also says it trained Argon to respect system boundaries and seek human confirmation when appropriate.

The announcement acknowledges that capable agents can attempt unauthorized shortcuts while trying to finish a task. Google says its monitoring systems are designed to detect serious boundary violations and block them. This makes agent behavior, permissions, and oversight a central part of the release alongside coding performance.

For development teams, the practical question is how reliably Argon completes work in their own repositories. A useful evaluation would measure accepted changes, regression rates, review effort, execution time, and total cost per completed task. Those results will show whether its broader capabilities translate into better software delivery.