Home · Technology · Oct 1 archive
Google Launches Gemini 4 Argon for Cyber Defenders
Confirmed
In Short: Google today unveiled Gemini 4 Argon, a new frontier AI model designed to enhance cybersecurity and complex workflows, rolling out to a select group of trusted cyber defenders through its Fairwind Program.
Built to handle deep reasoning in complex, long-horizon workflows, Argon is set to transform how Google works and builds internally.
Gemini 4 Argon has already been integrated into Google's internal workflows, aiding in specialized coding tasks, research, and writing.
According to Google, Argon has surpassed a published quantum algorithm optimization baseline by 40% within minutes and has identified data center memory optimizations that freed over 300 TiB of memory.
Gemini 4 Argon ties with OpenAI's GPT-6 Astra for the highest score on the CWE-bench, a test of how models can find and patch security vulnerabilities.
To ensure safety, Google has implemented systems to track Argon's chain of thought and stop it from performing when necessary, and is actively engaged in the US government's early-access framework for assessing cybersecurity risks.
Google is strengthening safeguards across four main areas before rolling Argon out broadly, including refusing harmful requests tied to cyber or CBRN attacks.
Wiz is already using Argon through its Scan for Good initiative, highlighting the model's strengths in specialized coding tasks and cybersecurity defense.
Gemini 4 Argon is Google's latest attempt to reassert itself in the race to build AI for high-stakes work, including security and coding.
What this adds
Gemini 4 Argon's expanded output token limit and its performance in complex workflows represent significant advancements over previous models.
Google's phased approach to releasing Argon includes gathering feedback from early testers to refine guardrails before broader release.
What's confirmed
- Today, we’re announcing our new frontier model, Gemini 4 Argon, which is rolling out to a set of trusted cyber defenders through our Fairwind Program.
- Gemini 4 Argon: our next era of frontier intelligence Gemini 4 Argon delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense.
- Built to sustain deep reasoning across complex, long-horizon workflows, Argon is fundamentally changing the way we work and build at Google.
- Argon will launch at an introductory price 1 of $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off input token price.
What's still developing
- Safely releasing frontier capabilities at this level requires a phased approach.
- We’ll continue to gather feedback from early testers as we iterate on guardrails before making Argon available to developers, enterprises, and consumers as soon as possible.
- It’s helping teams build faster and push the boundaries of engineering productivity and accelerating breakthroughs: To support Gemini 4 Argon’s capabilities across longer, more complex use cases, we are significantly expanding the model’s output token limit to an industry-leading 1M tokens, up from the previous 64K tokens.
- Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible.
- Google’s new Gemini 4 Argon equals GPT-6 Astra on the Artificial Analysis Intelligence Index at 60% of the Cost per Task with discounted prices.
- Google said the new model offers frontier performance in areas such as coding, cybersecurity defense, and enterprise knowledge work in domains like legal and finance.
- While Google has launched several updates to its Flash series of models, it has fallen behind the frontier to rival models from OpenAI and Anthropic in areas such as engineering and cybersecurity.
- It also said Argon has helped engineers free up more than 300 tebibytes of memory across its data centers without new hardware.
- Google said these safety measures also included precautions regarding how the model is given feedback to avoid teaching Argon to evade monitoring.
- The company previously promised to release a new frontier model, Gemini 3.5 Pro, in June.
- Google announced Gemini 4 Argon, its new frontier AI model, on September 30, 2026, in a post on Google’s official blog.
- Beyond security, Google said Argon raises the output token limit to 1 million tokens, up from the previous 64,000 tokens.