Skip to main content

Gemini 4 Argon: One Output, a Million Tokens

Google DeepMind's Gemini 4 Argon: A Million-Token Output Changes Everything

Imagine asking an AI to migrate an entire codebase, write a technical report, or automate a multi-step process—all in one go, without ever hitting a limit. That's exactly what Google DeepMind's new Gemini 4 Argon promises. Launched on September 30, its most jaw-dropping feature is the output limit: a staggering 1 million tokens in a single response. To put that in perspective, most cutting-edge models cap out around 128,000 tokens. Argon's jump isn't just incremental—it's a leap that could redefine what's possible in long-form, complex workflows.

Image

Built for the Long Haul

Argon isn't your average chatbot. Google is positioning it for "complex, long-term workflows"—think software engineering, legal analysis, financial modeling, and cybersecurity defense. And the benchmarks back it up. On DeepSWE v1.1, which tests long-term software engineering tasks, Argon scored a record 77.9%. It also topped the Vals Index (a weighted measure across finance, programming, law, and taxation), and achieved first place on Zapier's AutomationBench with 51.3%. For long video understanding, it hit 91.7% on LVBench—the highest yet. But it's not just about numbers; Argon can handle professional charts, extract details from lengthy videos, and perform cross-document actions with ease.

Security That Goes on the Offense

Cybersecurity is at the heart of this release. Argon is trained to autonomously discover, verify, and fix critical vulnerabilities. For trusted defenders and Google's internal teams, there's even a version with "no network barrier." In one case, security vendor Wiz used Argon through the Scan for Good project to uncover a serious vulnerability in a global hospital's medical software—something previous cutting-edge models had missed. On CWE-bench v1, which tests vulnerability repair, Argon ranked first with a 68% success rate.

Pricing and Access

During the introductory period, Argon costs $2 per million input tokens and $10 per million output tokens. After that, prices will double to $4 and $20, respectively. Before wide release, Google is beefing up defenses against abuse, prompt injection, misaligned reasoning chains, and sandbox escapes. Already, thousands of Google employees are using Argon to debug code, design algorithms, and perform large-scale code migrations. In one case, quantum algorithms were improved by 40% in minutes.

So, is Argon for everyone? Not quite. It's not aimed at casual chatting—it's for professionals tackling the hardest challenges. But if you're in software, law, finance, or security, this might just be the tool you've been waiting for.

Key Points

  • 1 million token output: Argon can generate up to 1 million tokens in a single response, far surpassing the typical 128K limit.
  • Top-tier performance: Record scores on DeepSWE v1.1 (77.9%), Vals Index (first), AutomationBench (51.3%), and LVBench (91.7%).
  • Security focus: Autonomously finds and fixes vulnerabilities; already caught a critical bug in hospital software.
  • Pricing: Introductory $2/$10 per million input/output tokens, later $4/$20.
  • Target audience: Professionals in software engineering, law, finance, and cybersecurity—not casual users.