Google Antigravity Launches Gemini 4 Argon for AI Pro Users
Google Antigravity announced today the launch of Gemini 4 Argon for paid Google AI Pro subscribers, setting up a separate weekly dedicated quota on top of the existing Gemini 3.8 Flash limits.
Gemini 4 Argon is Google's cutting-edge model, featuring a unified input-output window of 1 million tokens, optimized for complex software engineering, enterprise knowledge work, and cybersecurity tasks. Internal testing shows a 35% increase in agent code submissions and a 30% reduction in rollbacks after submission.
Previously, this model was primarily available to Google AI Ultra users and through enterprise APIs. Antigravity, as a priority development platform for agents, is now expanding access to the Pro subscription level due to high demand, with capacity being increased in phases.
Pro users, after receiving a separate weekly Argon quota, can use this model to handle complex workflows in the Antigravity desktop application and CLI, while continuing to use the 3.8 Flash limits. The API side maintains an introductory price until November 30, at $2 per million input tokens and $10 per million output tokens.
Internally, over 100,000 Alphabet users utilize Argon through an internal version of Antigravity each month to address production codebase issues, with the Chrome team having used it for large-scale refactoring of Chromium.
Paid subscribers gain access to higher-tier models, with event-driven demand reflecting the computational needs of cutting-edge models and subscription tiers. The funding flows towards Google AI subscriptions and API usage, benefiting developers and enterprises willing to pay for Pro or higher tiers, while putting pressure on free and lower-tier users and infrastructure with limited computational resources.
Source: Public Information
ABAB AI Insight
Google DeepMind developed the Gemini series, with Gemini 4 Argon announced at the end of September 2026 as a next-generation cutting-edge model, initially targeting internal and specific cybersecurity users, and later expanding to Ultra and enterprise, now further opening to the Pro tier. Antigravity is Google's priority software development platform for agents, integrating multi-agent workflows and code execution.
In terms of capital flow, Google monetizes high computational cost cutting-edge models through subscription tiers and API pricing, with Pro users receiving a separate weekly quota instead of sharing existing limits. The motivation is to balance peak demand with capacity while tying higher capabilities to paid tiers; the behavior is characterized by phased capacity expansion and dedicated quotas to avoid resource consumption by free tiers.
Comparable cases include OpenAI's subscription tiers and rate limits for the GPT series, as well as Anthropic's tiered access to Claude. The industry is currently transitioning from internal validation of cutting-edge models to commercializing paid tiers, with Google binding agent coding capabilities to subscriptions through Antigravity.
Essentially, this represents a transfer of pricing power: cutting-edge model capabilities are allocated based on subscription tiers and dedicated quotas rather than unified access, with higher-tier models tied to payment. The mechanism is driven by a mismatch between computational costs and demand, allowing the company to control resource consumption through independent quotas and introductory pricing while concentrating the highest capabilities among paying users, influencing developers' choices regarding platforms and models.
ABAB News · Cognitive Laws
Cutting-edge capabilities are allocated by subscription tier rather than open on demand
Independent quotas serve as pricing tools under capacity constraints
The cost of agent coding determines access levels