AI NexusAI Nexus
Product update

Google announces Gemini 4 Argon with 1M-token output and Fairwind access

Event time 2 independent sourcesEditorial score 80/100Updated here

The short version

Google introduced Gemini 4 Argon, a frontier model for complex real-world coding, enterprise knowledge work, and cyber defense with an output limit raised to one million tokens. It is now rolling out to trusted cyber defenders through the Fairwind Program, with an introductory price of $2 per million input tokens and $10 per million output tokens.

What changed

Compared with the previous 64K output ceiling, Gemini 4 Argon raises the output token limit to one million and initially opens to cyber defenders on a restricted basis.

What it means for you

Users who care about coding and long-horizon tasks should track DeepSWE v1.1 and Vals Index results, but Google is the source; wait for official access before relying on them.

How to use it: watch Google\u2019s official pages for Argon availability; until it is generally available, compare current Gemini APIs with Claude and Codex tools.

Read this with caution: benchmarks such as DeepSWE and Vals Index come from Google\u2019s own blog and are vendor claims; the model has not been broadly opened to developers, enterprises, or consumers yet.

Fact check

  • VerifiedGoogle 在博客中宣布 Gemini 4 Argon 是面向真实软件工程、法律和金融等企业知识工作、以及网络防御的新前沿模型Evidence
  • VerifiedGoogle 称 Gemini 4 Argon 将输出 token 上限提升到 100 万，此前为 64KEvidence
  • Verified目前先通过 Fairwind Program 向受信任的网络防御者开放，之后才逐步向开发者、企业和消费者提供Evidence
  • Verified方案公布首发价格是输入每百万 token 2 美元、输出每百万 10 美元，缓存输入价格为首发输入价的 95% offEvidence

Coverage timeline

  1. Primary sourceGoogle DeepMind
    Google DeepMind
  2. Primary sourceGoogle
    Google