Google widened access to its Gemini 4 Argon model for paying users and developers after testing it with cybersecurity teams. The rollout includes discounted token rates that last until November 30.
Google has released its Gemini 4 Argon artificial intelligence model to developers and paying subscribers worldwide, expanding availability beyond an initial test group, CNET reported. The update allows enterprises and consumer subscribers to use the system across several Google services after the company restricted early access to cybersecurity specialists and government partners.
Why the release matters for everyday work
In the consumer Gemini app, Argon delivers responses faster and cuts down on hedging, repetition, and clichés, according to Josh Woodward, vice president of product for Gemini and Google Labs. Users can read generated responses while the model continues processing longer requests, and the app now lets people switch between chat and tasks more easily, Engadget reported. Google is also putting the Antigravity agentic coding interface into its macOS application.
The model is built to manage work in finance, coding, software engineering, creative writing, and defensive cybersecurity. Google reported that its own teams used Gemini 4 internally to optimize Python code, spot and fix security vulnerabilities, and run open Gemma models three times faster. In robotics experiments, engineers used the model to help a humanoid machine follow real-world physics and interpret its physical surroundings.
Independent benchmark measurements also show performance gains against rival systems. Testing by Artificial Analysis found that Gemini 4 Argon equals OpenAI’s GPT-6 Astra on its composite Intelligence Index. At Google’s introductory rates, the system achieved that score at 60 percent of Astra’s cost per task.
What Google did before expanding access
DeepMind announced Gemini 4 Argon a few weeks ago before restricting initial rollouts. Earlier this month, Google gave access only to cybersecurity defenders and members of its Fairwind Program, an initiative for governments and trusted partners that require advanced defense capabilities. The company had planned to introduce a Gemini 3.5 Pro version, but it abandoned that release and moved straight to Argon.
The launch comes after shifts in Google’s competitive position. Google led among AI developers for most of 2025, but advancements by competitors left it mostly behind for this year as it competes with OpenAI’s GPT-6 Astra and Meta’s Muse Wonder. Google previously gained attention with Nano Banana, an image generator that beat rivals but drew anger from creators. The company has now shifted toward marketing Argon as a tool for technical workers, math, and scientific research, including robotics and nuclear fusion. Other tech firms have adopted similar messaging around goals such as cancer treatments, as critics raise concerns over data center construction and invasive automated agents.
What happens next with pricing and safety
Subscribers in the United States can access Argon through Google AI Pro for $20 a month or the Ultra plan for at least $100 a month. Paying users can run the model in the Gemini app and through AI Mode in Search. Developers can use the system through the Gemini API, Google AI Studio, Android Studio, and the Antigravity agentic development platform.
Introductory token fees are set at $2 per million input tokens and $10 per million output tokens. Those rates remain in place until November 30. Once that deadline passes, prices double to $4 per million input tokens and $20 per million output tokens.
Safety reviews will continue under scrutiny after rising concern over the past year about AI safety and doomsday scenarios. Google took part in a voluntary White House review program for new models, and the company stated that Gemini 4 passed inspections under its safety framework. The public does not have visibility into that federal review process.
