Google releases Gemini 4 Argon AI model for paying subscribers

In short: Google has launched Gemini 4 Argon, a new version of its Gemini AI that it says can match top competitors, and it is starting with paying users.

What happened

Google released Gemini 4 Argon on October 13, 2026. Argon is the underlying AI model, meaning the “brain” that powers features like chat, writing help, and coding tools.

Google says Argon is better at complex, multi step work, and it repeats itself less. The company also says it avoids the filler language that can show up in AI chatbots, like cautious hedging and overused phrases.

According to Google, Argon performs better than rival models such as OpenAI’s GPT-6 Astra and Anthropic’s Claude Fable 5.1 and Claude Opus 5.5 on standard tests. These tests are often called “benchmarks,” which are like timed exams used to compare models on tasks like coding, cybersecurity, science, finance, and law.

Google says it is already using Argon inside the company. Examples include finding and fixing security holes, reducing the computing power needed to run YouTube’s comments systems, and helping debug a drone physics simulator used by Wing, Alphabet’s drone delivery group.

Argon is rolling out to higher tier Pro and Ultra subscribers in the Gemini app and AI Mode in Google Search. It is also coming to Ultra users on Google’s Antigravity platform, which is aimed at building “agent” tools (AI that can take actions for you, like a digital helper that follows a checklist). Google did not say when, or if, it will come to free users.

Why it matters

This launch gives Google a new option for people and businesses comparing paid AI tools, especially on price. Google is starting Argon at $2 per million input tokens and $10 per million output tokens, with prices set to double after November, and “tokens” are small chunks of text that AI systems count like meter units.

Source: The Verge AI