Google Opens Gemini 4 Argon to the Public
Google Opens Gemini 4 Argon to the Public
Google is making Gemini 4 Argon broadly available today. The rollout opens access through the paid Gemini API, Google AI Studio, the Antigravity agentic development environment, and eligible Gemini Enterprise subscriptions. Google AI Pro and Ultra subscribers can also reach the model through the Gemini app and AI Mode in Google Search.
The release marks the first time Argon is available outside a small group of vetted cybersecurity professionals. Google announced the model on September 30 but held general access back while conducting additional safety testing and participating in a voluntary U.S. government pre-release review process.
What Argon Is
Gemini 4 Argon is Google DeepMind's current flagship model. It is built for coding, complex knowledge work, and cyber defense applications. The headline capability is a 1 million token output limit. That figure represents a significant increase over prior Gemini generations and most competing models.
Large output windows matter most for tasks that require generating long documents, complex codebases, or extended reasoning chains in a single pass. Where most models require breaking large tasks into sequential calls, a million-token output ceiling allows a single request to carry more of that work through to completion.
Google has also been running Argon on its own internal engineering problems ahead of today's release. The company says improvements from that work have been incorporated into the public Agent Development Kit Python codebase.
A 1 million token output limit changes the unit of work. Tasks that required multiple sequential calls can now complete in one.
Access and Pricing
Access to Argon is tiered. Developers and enterprises reach the model through the paid Gemini API and Google AI Studio. Antigravity is available to enterprise subscribers and to individual Google AI Ultra subscribers. Google AI Pro subscribers can access Argon in the Gemini app and AI Mode in Search, but Antigravity on an individual plan requires the more expensive Ultra tier.
Introductory API pricing starts at $2 per million input tokens and $10 per million output tokens. Google has confirmed those rates will rise — to $4 and $20 respectively — when the introductory period ends. Batch processing is available at $1 per million input tokens and $5 per million output tokens. Cached input carries a 95 percent discount.
The Path to Today
Argon's September 30 announcement came paired with an unusual access restriction. Rather than opening to developers immediately, Google gave first access to participants in its Fairwind Program — a group of approved cybersecurity defenders. The stated rationale was continued safety testing and a voluntary government pre-release review.
Today's broader rollout follows that 13-day window. Google says it intends to expand access further to developers, enterprises, and general consumers but has not announced a specific timeline for those additional tiers.
Argon's arrival also reflects a notable shift in Google's product roadmap. The company had been expected to release a Gemini 3.5 Pro update before a next-generation model. It did not. Google DeepMind chief Koray Kavukcuoglu confirmed in late September that the team stepped back from Pro-tier iterations to focus on Flash models, then moved directly to Gemini 4.
Google skipped Gemini 3.5 Pro entirely, moving its flagship model lineup from 3.1 Pro straight to Gemini 4 Argon.
What Comes Next
Google has not announced a successor to Argon or a timeline for a Gemini 4 Flash variant. Kavukcuoglu described the current release as an early post-training output, with continued fast-paced iteration to follow. The company has said it expects to keep improving the Gemini 4 family but has not committed to specific milestones or dates.
Gemini 4 Argon is available today through the Gemini API and Google AI Studio at ai.google.dev.
Aaron Rose is a software engineer and technology writer covering system architecture, cloud platforms, and AI policy.