Skip to main content

Google introduces Gemini 4 Argon with staged cyber rollout

13 OCTOBER 2026·3 MIN READ·5 SOURCES

Google has introduced Gemini 4 Argon, which it describes as a new frontier model for complex workflows in software engineering, enterprise knowledge work and cybersecurity defense. Its initial rollout is aimed at trusted cyber defenders, while Google gathers feedback and refines safeguards.

Google introduces Gemini 4 Argon with staged cyber rollout

Key takeaways · 3

  • 01

    Argon’s first rollout is through Google’s Fairwind Program for trusted cyber defenders, with access planned to broaden after testing and guardrail iteration.

  • 02

    The introductory rates are $2 per million input tokens and $10 per million output tokens, with cached input tokens priced 95% below the input rate.

  • 03

    Google says Argon can generate validated code fixes after identifying vulnerabilities; broader availability is planned, but Google has not given a specific date in the evidence provided.

A staged security-first release

Google says Argon is built to sustain deep reasoning across complex, long-horizon workflows.[1] The company initially rolled it out to trusted cyber defenders through its Fairwind Program, and says it will use early-tester feedback to iterate on guardrails before making the model more broadly available.[1] Google also says it is participating in the U.S. government’s voluntary process for pre-release model access.[1] It plans to make Argon available to developers, enterprises and consumers as soon as possible, but the evidence does not provide a date for that wider release.[1]

Pricing and access details

Google set introductory pricing at $2 per million input tokens and $10 per million output tokens.[1] Cached input tokens are priced at a 95% discount to the input-token rate.[1] The Verge reported that Argon’s introductory token prices were around one-fifth the prices of OpenAI’s GPT-6 Astra and Anthropic’s Fable 5 models.[3] It also reported that Argon’s prices would double when the introductory period ended at the end of November.[3] The Verge said Argon was rolling out in the Gemini app and Search’s AI mode to Google Pro and Ultra subscribers, and on Antigravity to Ultra subscribers.[3] It said it was unclear whether or when Plus subscribers or free users would get access.[3]

Capabilities and reported results

Google describes Argon as delivering frontier performance across software engineering, enterprise knowledge work and cybersecurity defense.[1] Google DeepMind says Argon can generate validated code fixes automatically after finding vulnerabilities, and ties for first on CWE-Bench with a score of 68%.[4][5] Google also calls it its most resilient model against indirect prompt injections.[5] Google says thousands of its employees highlighted strengths in specialized coding, deeper research and writing quality.[1] The Verge reported that Google said Argon outperformed GPT-6 Astra, Claude Fable 5.1 and Claude Opus 5.5 across benchmarks including coding, cybersecurity, science, finance and law.[3] These are company claims and reported benchmark comparisons, not evidence here of results in every organization’s workflows.

Early use cases and partner expansion

Google DeepMind says Argon is expanding to hundreds of additional Fairwind Program partners.[5] Cybersecurity company Wiz used early access to Argon to protect critical healthcare infrastructure and hundreds of thousands of patient records, according to Google.[4] The Verge reported that Google used Argon to halve the computing power needed for YouTube’s comments section and to debug a drone-physics simulator for its delivery company Wing.[3] These examples span security, infrastructure and engineering tasks, but they do not establish that the same outcomes will transfer to other deployments.

Argon combines a security-focused initial rollout with a published token price and a stated plan for broader access. Professionals evaluating it can use those details to frame a pilot, while treating benchmark and early-use claims as points to validate against their own tasks.

Why it matters
Story quiz

Test yourself on this story — 1 question.

Create a free account to take the quiz, earn XP, and get a daily session built for your industry.

Take the quiz

How this developed

  1. 13 October 2026

    Google introduces Gemini 4 Argon with staged cyber rollout

Sources

AI fluency, one session a day, built for your work.