Google's Gemini 4 Argon Launches for Paying Users, with Coding Features Restricted to Top Tier

Key points

• Gemini 4 Argon is now available in the Gemini app and AI Mode in Search for Google AI Pro ($20/month) and AI Ultra (starting at $99/month) subscribers.
• Advanced agentic coding features in the Antigravity platform are exclusively available to AI Ultra subscribers, while Pro users can only use basic code generation in Canvas mode.
• Developers can access the model via the API at introductory rates of $2 per million input tokens and $10 per million output tokens, with a 95% discount for cached tokens.
• These introductory API prices expire on November 30, after which rates will double to $4 per million input tokens and $20 per million output tokens.
• The model is not available to free users, who are limited to the Flash-Lite model, and Google has not confirmed when or if broader access will be granted.

Background

Google previously delayed the release of its 3.5 Pro model, opting instead to launch Gemini 4 Argon after initial testing with the Fairwind Program, a group of government and trusted partners. The model was first introduced in late September 2026, with Google citing the need for cautious rollout due to potential misuse. This launch follows recent moves by Google to restrict free-tier access to advanced AI features, such as limiting Flash models and expanding AI design tools to paid plans.

How outlets are covering it

Engadget and Ars Technica both emphasize the tiered access model, noting that while Pro subscribers can use Argon for general tasks, only Ultra subscribers can access the more compute-intensive Antigravity coding platform. Ars Technica highlights that Google has not clarified when or if broader access will be granted, while Engadget notes that the model's introductory pricing is a temporary measure. Google's official blog focuses on the model's performance in complex tasks, such as coding, finance, and cybersecurity, and highlights its efficiency gains in internal projects, such as optimizing Chrome's codebase and improving quantum chemistry software. Ars Technica also points out that Google has kept Gemini 3.1 Pro available for users who need a less expensive alternative to Argon.

Why it matters

The release of Gemini 4 Argon signals Google's continued push to compete with OpenAI and Anthropic in the frontier AI market. By restricting advanced features to higher-tier plans, Google is likely aiming to maximize revenue from its most capable models. The model's performance in complex tasks, such as coding and cybersecurity, could have significant implications for industries that rely on AI for automation and problem-solving. Additionally, the temporary pricing structure may encourage developers to adopt the model before prices increase, potentially shaping the future of AI development and deployment.

What to watch

Google will likely monitor the adoption of Gemini 4 Argon and may adjust pricing or access tiers based on demand. The company may also expand access to the model for lower-tier subscribers or introduce new features to differentiate its offerings. Developers and enterprises will need to evaluate whether the introductory pricing is worth the potential cost increase after November 30. Additionally, Google may continue to refine the model's safety measures and performance in response to feedback from users and partners.