Google Unlocks Gemini 4 Argon for Paying Users, Restricts Coding Features to Top Tier

Key points

• Gemini 4 Argon is now available in the Gemini app for AI Pro and AI Ultra subscribers, but advanced agentic coding workflows in Antigravity are restricted to AI Ultra users only.
• API pricing is temporarily set at $2 per million input tokens and $10 per million output tokens, with a 95% discount for cached tokens, until November 30, 2026.
• Google claims Argon has improved internal efficiency, including a 45% reduction in server load for YouTube comments and the identification of critical bugs in Project Wing dating back to 2018.
• The model ranks first on the Vals Index for critical economic sectors and demonstrates significant improvements in embodied reasoning for robotics tasks.
• Free users remain restricted to Flash-Lite, as Google has not announced plans to bring Argon to lower-tier or free plans.

Background

Google initially delayed the full release of Gemini 4 Argon due to safety concerns regarding its advanced capabilities, particularly in cybersecurity and coding. Earlier in October, the model was restricted to vetted cybersecurity partners through the Fairwind Program. This marks the first broad commercial rollout, following previous restrictions on Flash models for free users.

How outlets are covering it

Ars Technica emphasizes the restrictive nature of the rollout, highlighting that even Pro subscribers lack access to full coding capabilities in Antigravity, which are reserved for the more expensive Ultra tier. In contrast, Google’s official blog focuses on the model’s performance and enterprise impact, citing specific improvements in code submission rates and reductions in rollbacks. While Ars notes the temporary nature of the discounted API rates, Google’s communications frame the pricing as a standard introductory offer for developers. Both sources agree on the tiered access structure but differ in tone regarding the exclusivity of the coding features.

Why it matters

The tiered release strategy signals Google’s intent to monetize its most advanced AI capabilities through higher subscription tiers, potentially creating a divide between casual users and professional developers. The aggressive introductory API pricing may attract developers, but the doubling of rates in November could impact long-term integration costs. This move also highlights the growing importance of agentic coding and spatial reasoning in AI development, as Google positions Argon as a tool for complex, multi-step tasks across various industries.

What to watch

Developers should evaluate their workflows before November 30, 2026, to determine if the introductory API rates justify the potential cost increase. Users on lower-tier plans may need to upgrade to AI Ultra to access advanced coding features. Google has not announced when or if Argon will become available to other subscription tiers, leaving uncertainty about future accessibility.