CALIFORNIA / RankWire.AI / – Google has introduced Gemini 4 Argon, its latest flagship artificial intelligence system designed for handling intricate professional tasks. The company unveiled this model on Sept. 30, establishing it as the cornerstone of its Gemini 4 series. Argon is aimed at areas such as software development, finance, legal analysis, and cybersecurity defense. Google highlighted that the model is capable of maintaining deeper reasoning over extended, multi-step processes. Currently, access is limited to a select group of cybersecurity defenders participating through its Fairwind Program.
The Gemini 4 Argon model increases the maximum token output to 1 million, a significant jump from the previous cap of 64,000 tokens. This larger limit enables the system to process lengthy tasks without interruption within a single session. The initial API pricing has been set at $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95%. Following the initial rollout, costs are expected to rise to $4 and $20, respectively.
According to Google, thousands of its employees are already leveraging Argon for tasks including coding, research, and content creation. The model is also being used internally for data center memory optimization and large-scale code migrations. One such project involved deploying Argon agents for migrating C and C++ codebases to Rust. Another utilized agents for memory profiling across Google’s data centers. The company reported that these optimizations freed over 300 tebibytes of memory, with additional savings identified through similar efforts.
Enhanced abilities for sophisticated professional applications
Google shared that Gemini 4 Argon scored 77.9% on DeepSWE v1.1, a benchmark evaluating long-term software engineering performance. The model also demonstrated strong results in finance, legal work, and automation testing. It supports multimodal reasoning alongside coding and enterprise tasks. Developed by Google DeepMind as part of the larger Gemini family, Argon’s increased output capacity enables it to handle extended workflows involving multiple reasoning and execution steps. Google emphasized that this capacity is crucial for complex, multi-stage professional projects.
Another key aspect of Argon’s deployment is its application in cybersecurity. Google states that the model can identify, verify, and patch software vulnerabilities within controlled defensive environments. Through its Scan for Good initiative, Wiz is utilizing Argon to detect security risks in public infrastructure. Google reported that Argon achieved a score of 68% on CWE-bench v1, a benchmark for vulnerability mitigation. Access to this model is currently limited to select defenders working without the usual cyber guardrails, for approved security activities.
Gradual release precedes wider availability of Gemini 4
A specific launch date for the general public release of Gemini 4 Argon has not been announced by Google. The company explained that it is employing a phased rollout approach, gathering feedback from early users. Additionally, Google is participating in a voluntary pre-release process with the U.S. government for early model access. The broader availability is expected to include developers, enterprises, and consumers later on. Initial rollout will focus on paid API users and Google AI Ultra subscribers, though no official date has been provided for these groups.
Google also clarified that it has no plans to release Gemini 3.5 Pro, which was initially scheduled for June. As a result, Gemini 4 Argon remains the company’s latest flagship model tailored for demanding reasoning and professional use cases. The company continues to offer other Gemini variants catering to different performance and budget needs. Argon distinguishes itself with higher output capacity, advanced coding features, and specialized cybersecurity capabilities. Currently, its availability is limited to trusted testers and carefully selected security partners.