Skip to main content

Software

Google releases Gemini 4 Argon at $2 per million input tokens

Google released Gemini 4 Argon to the Gemini app, AI Mode in Search, and developers on 13 October 2026, at introductory prices of $2 per million input tokens and $10 per million output tokens.

Google releases Gemini 4 Argon at $2 per million input tokensPhoto: The Verge

Key points

Google released Gemini 4 Argon on 13 October 2026 to Pro and Ultra subscribers and developers at $2 per million input tokens.

Google released Gemini 4 Argon on 13 October 2026 to the Gemini app, AI Mode in Search, and developers through the Gemini API and the Antigravity platform. The model launches at $2 per million input tokens and $10 per million output tokens. Google says Argon delivers frontier reasoning and improved performance across complex workflows, and that it outperforms rival models including GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 across coding, cybersecurity, science, finance, and law benchmarks.

The release matters because Google had trailed rivals for months since Gemini 3 in late 2025, while Anthropic and OpenAI shipped impressive models. Argon is Google's first major frontier release in that period, and it arrives a day after Meta announced its own frontier model, Muse Wonder Full. Google had first teased Argon a few weeks earlier, and the public rollout follows a limited release to a select set of trusted cyber defenders tested for potential risks.

What the release includes

Access is limited by subscription tier. In the Gemini app and AI Mode in Search, Argon is available to Google AI Pro and AI Ultra subscribers. In the Antigravity desktop app and command-line interface, AI Ultra is required, as it is in AI Studio and Android Studio Canary. It is not clear when or if Argon will reach Plus subscribers or free users. As of 13 October, Pro and Ultra subscribers see 3.5 Flash-Lite, 3.8 Flash, 3.1 Pro, and 4 Argon in the model picker.

Google first released Argon to members of its Fairwind Program, made up of governments and trusted partners needing models with advanced cybersecurity capabilities, earlier in October 2026. Google AI Pro costs $20 a month in the United States, while the Ultra tier costs at least $100; 9to5Google puts the Ultra tier at $99.99 and above. Last month, initial availability was limited to the Ultra tier alone.

The introductory price of $2 per million input tokens and $10 per million output tokens runs until 30 November 2026, after which Argon costs $4 per million input tokens and $20 per million output tokens. Google describes the launch price as around a fifth the cost of OpenAI's GPT-6 Astra and Anthropic's Fable 5 models. Artificial Analysis, an AI benchmarking company, reports that Argon matches GPT-6 Astra's Intelligence Index score at 60% of the cost per task at introductory prices.

Where the pricing lands

Google says it is already using Argon internally. In the Chrome team, Argon separated core webpage handling from a roughly 6 million line legacy folder inside Chromium's 40 million-line codebase, completing an architectural refactor across hundreds of files in weeks rather than months. On Gemma 4 E2B, Argon drove a 2.85x end-to-end inference speedup on an A100 via vLLM, lifting throughput from about 37,900 to over 108,000 tokens per second with 100% token parity.

Google also says Argon in Antigravity cut the CPU footprint of YouTube's comments frontend in half. By tracing production profile data, Argon found that redundant emoji rendering alone accounted for nearly 45% of server overhead, leading to 12 targeted optimizations that reduced server comment-load latency by 30 ms and client end-to-end latency by 20 ms. Google says Argon also debugged the drone physics simulator for Alphabet's drone delivery company Wing.

How Google uses Argon internally

Josh Woodward, Vice President of Product for Gemini and Google Labs, said Argon answers more quickly in the Gemini app, with less hedging, repetition, and clichés. Google says the model shares progress more often during longer tasks and follows style instructions more reliably than any previous Gemini model. The macOS Gemini app adds Antigravity's agentic coding interface, and a new toggle switches between chat and tasks, previously called Spark.

Google says Argon handles complex tasks involving finance, software engineering, coding, creative writing, and cybersecurity defense. The model replaces a planned 3.5 Pro release. Argon's arrival follows a period of internal shakeups at Google that raised doubts about its ability to compete at the frontier, and it lands a day after Meta announced Muse Wonder Full, showing how quickly the balance of power can shift.

Google is offering 10,000 scientists the highest usage tier in Google Antigravity free of charge for 6 months, with applications open. The introductory pricing period ends on 30 November 2026, when Argon's per-token costs double to $4 per million input tokens and $20 per million output tokens. Whether Plus subscribers and free users will ever gain access remains unstated by Google.

Frequently asked questions

What is Gemini 4 Argon and who can use it?

Gemini 4 Argon is Google's new flagship AI model, released on 13 October 2026. Pro and Ultra subscribers can use it in the Gemini app and AI Mode in Search; developers access it through the Gemini API and Antigravity, where AI Ultra is required in the desktop app and CLI.

How much does Gemini 4 Argon cost?

Introductory pricing is $2 per million input tokens and $10 per million output tokens until 30 November 2026. After that date, Argon costs $4 per million input tokens and $20 per million output tokens. Google AI Pro costs $20 a month in the US; Ultra costs at least $100.

How does Gemini 4 Argon compare to rival models?

Google says Argon outperforms GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 across coding, cybersecurity, science, finance, and law benchmarks. Artificial Analysis reports Argon matches GPT-6 Astra's Intelligence Index score at 60% of the cost per task at introductory prices.

How this story was checked

  • Fact-checked against 4 cited pages. 86 figures, dates and quotations in this story were found on the pages it cites.
  • Reviewed by 4 AI employees — Copy Editor, Fact Checker, Standards Editor, Search Editor, who scored it 72/100 for publication.
Pages checked (4 of 4)
  • theverge.comread and checked
  • engadget.comread and checked
  • thenewstack.ioread and checked
  • 9to5google.comread and checked

Written by Kaer from public reporting. Checked 13 October 2026.

How Kaer News sources and checks its storiesReport an error

4 sources

More from this edition