Google Launches Gemini 4 Argon in Limited Release After Delays

Google DeepMind announced Gemini 4 Argon on September 30, 2026, its new top-tier artificial intelligence model and the first of the Gemini 4 generation, following months of reported delays. The model is currently available only to a restricted group of users, including vetted cybersecurity professionals, as the company manages a phased rollout.

What Happened

Alphabet Inc.'s Google DeepMind publicly unveiled Gemini 4 Argon on September 30, 2026, positioning it as the company's most capable frontier model to date. The announcement came after a period Reuters described as "months of delays." The model is not yet broadly accessible. According to reporting from VentureBeat and Technology.org, the initial release is limited, with cybersecurity defenders among the first authorized user groups.

The model supports up to one million output tokens, according to MarkTechPost, targeting use cases including long-horizon software development, knowledge work, and cyber defense tasks.

Benchmark Claims

Google says Gemini 4 Argon leads rival systems on several benchmarks. Technology.org reported the model tops 12 of 18 benchmarks the company disclosed. VentureBeat described those benchmarks as "enterprise-relevant," and multiple outlets reported Google framed the results as reclaiming a performance lead over OpenAI and Anthropic.

Benchmark methodology and independent verification details were not available in the wire reports reviewed. Google's benchmark claims are self-reported as part of the product announcement.

Limited Rollout and Early Problems

Despite the announcement, the model is not in wide release. Gizmodo reported that Google is already encountering problems with the model, though the specific nature of those issues was not fully detailed in available wire copy. The Los Angeles Times reported that Google is "grappling with" challenges as part of the launch, without specifying their scope.

The phased rollout means most Google and Google Cloud users do not yet have access to Gemini 4 Argon. VentureBeat confirmed the limited release status in its coverage of the announcement.

Background

Google DeepMind is the artificial intelligence research division of Alphabet Inc., formed through the merger of Google Brain and DeepMind in 2023. The Gemini model family is Google's primary large language model line, competing directly with OpenAI's GPT series and Anthropic's Claude models.

The AI model market has seen rapid successive releases from all three companies across 2025 and 2026. Google's announcement comes as the competitive landscape has intensified, with both OpenAI and Anthropic releasing updated flagship models in recent months. The INDmoney report described the current environment as a race among "billion-user platforms" to expand AI capabilities across enterprise and consumer applications.

Separately, the United States Federal Trade Commission has opened an investigation into OpenAI, Anthropic, and other AI companies to assess potential risks associated with their products, according to a report from Informat.ro. That investigation covers the broader AI industry and is not specific to Google or Gemini 4 Argon.

Target Applications

Google has identified three primary use cases for Gemini 4 Argon at launch: software engineering tasks requiring extended context, enterprise knowledge work, and cybersecurity. The cyber defense focus is reflected in the initial access restrictions, which prioritize vetted security professionals, according to Technology.org.

The one-million output token capacity is presented by Google as a differentiator for tasks requiring sustained, long-form generation, such as producing or reviewing large codebases.

What Comes Next

Google has not publicly announced a timeline for broader general availability of Gemini 4 Argon, and the company's phased rollout approach means access expansion dates remain unconfirmed as of the date of this report.