Google Opens Gemini 4 Argon AI Model to Subscribers, Catches Up With GPT-6 Astra

Google has made its Gemini 4 Argon artificial intelligence model available to more users. AI Pro and AI Ultra subscribers can use the model in the Gemini app and the search engine’s AI mode. Developers can access it through the Gemini API and the Antigravity platform. The model was initially provided to governments and trusted partners that needed advanced cybersecurity tools.

Gemini 4 Argon is designed for complex tasks such as finance, software engineering, coding, creative writing, and cybersecurity. Google moved directly to the new version without releasing its previously planned Gemini 3.5 Pro model. The model provides shorter answers, with less repetition and fewer formulaic phrases. For lengthy tasks, it does not leave users waiting for the result and shows them what stage the work has reached.

In Artificial Analysis tests, Gemini 4 Argon achieved the same overall score as OpenAI’s GPT-6 Astra model. Based on initial pricing, the cost of each task was approximately 60% of the amount charged for the competing model. An AI Pro subscription costs 20 dollars per month (approximately 34 manats). An AI Ultra subscription costs at least 100 dollars per month (approximately 170 manats).

For developers, the API costs 2 dollars (approximately 3 manats) per 1 million input tokens. Here, a token refers to a small segment of the text sent to the model. The 1 million output tokens generated by the model cost 10 dollars (approximately 17 manats). These prices will remain in effect through November 30, 2026. After that, the input price will double to 4 dollars (approximately 7 manats). The output price will also rise to 20 dollars (approximately 34 manats).

Antigravity is a software development platform that plans tasks on its own. Version 2.0 was demonstrated at the I/O 2026 event. The platform previously operated as a separate app for Mac, Windows, Linux, and Googlebook. Antigravity has now been integrated directly into the macOS app. Users can switch from the chat section to coding in the same app and create apps and web projects.

The ability to switch between the Chat and Task sections has also been added to the macOS app. Task is the new name of the section previously called Gemini Spark. The Scheduled page displays tasks and prescheduled operations in one place. To access features, users now type the @ symbol instead of the previous slash command. Connected apps are now called Connectors. The model list includes 3.5 Flash-Lite and 3.8 Flash. The other options are 3.1 Pro and 4 Argon.

In internal tests, the Chrome team reduced the time required to update 40 million lines of Chromium code from months to weeks. The processing speed of the Gemma 4 E2B model increased approximately 3 times without any loss of quality. The 12 changes made to YouTube comments reduced server response latency by 30 ms. Overall on-device latency also decreased by 20 ms. Google will also give 10,000 scientists the highest-tier Antigravity plan free of charge for 6 months.