CALIFORNIA / RankWire.AI / – Google has introduced Gemini 4 Argon, its latest flagship AI model designed for handling complex professional tasks. The company announced this model on Sept. 30, marking it as the cornerstone of its Gemini 4 series. The Argon model is tailored for software development, financial, legal work, and cybersecurity defense. Google highlighted that the model can maintain deeper reasoning over extended, multi-step processes. Currently, access is limited to a select group of cybersecurity defenders through its Fairwind Program.
With Gemini 4 Argon, the maximum output limit has been increased to 1 million tokens from the previous 64,000 tokens. This enhancement allows the model to process longer tasks within a single sequence. Initial API pricing is set at $2 per million input tokens and $10 per million output tokens. Input tokens stored in cache benefit from a 95% discount on the input price. Following the introductory phase, prices will be adjusted to $4 and $20, respectively.
According to Google, thousands of employees are already utilizing Argon across specialized coding, research, and writing activities. Internal teams have also used the model to optimize data center memory and facilitate large-scale code migrations. One project involved Argon agents assisting with C and C++ to Rust migrations, while another focused on memory profiling within Google’s data centers. The company stated that these optimizations have freed over 300 tebibytes of memory, with additional savings identified through similar efforts.
Enhanced Capabilities for Complex Professional Tasks
Google reported a 77.9% score for Gemini 4 Argon on DeepSWE v1.1, a benchmark assessing long-term software engineering performance. The company also emphasized results in finance, legal work, and automation testing. Argon supports multimodal reasoning along with coding and enterprise-related tasks. Developed by Google DeepMind as part of the broader Gemini family, the model’s increased output capacity enables it to better manage extended workflows that involve multiple consecutive reasoning and execution steps.
Cybersecurity remains a key focus during the initial deployment phase. Google claims Argon can identify, verify, and patch software vulnerabilities within controlled defensive environments. Through its Scan for Good initiative, Wiz is employing Argon to detect security vulnerabilities in public infrastructure. Google noted that Argon achieved a 68% score on CWE-bench v1, a standard benchmark for vulnerability remediation. Selected security defenders are being granted access without Google’s standard cyber safeguards for specific defensive tasks.
Gradual Launching Before Broader Access to Gemini 4
Google has not specified a date for the widespread public release of Gemini 4 Argon. The company explained that it is adopting a phased rollout, actively gathering feedback from early users. Additionally, Google is participating in a voluntary pre-release program with the U.S. government for model access. The broader availability, according to Google, will eventually include developers, enterprises, and consumers. The initial rollout will target paid API users and Google AI Ultra subscribers, though no exact launch date has been provided for these groups.
Furthermore, Google announced it has no plans to release Gemini 3.5 Pro, which was originally scheduled for June. Consequently, Gemini 4 Argon remains the company’s latest flagship model for demanding reasoning and professional workloads. Google still offers other Gemini versions tailored to different performance and cost needs. Argon stands out with its larger output limit, advanced coding capabilities, and specialized cybersecurity features. At present, the model is accessible only to trusted testers and selected cybersecurity partners.