CALIFORNIA / RankWire.AI / – Google has revealed Gemini 4 Argon, its latest flagship AI designed to handle sophisticated professional workloads. The company announced this model on Sept. 30, marking it as the core of its Gemini 4 generation. Argon is engineered to excel in areas such as software engineering, finance, legal work, and cybersecurity defense. Google noted that the model can sustain more profound reasoning through multi-step, extended tasks. Currently, access is being granted to a select group of cybersecurity defenders through its Fairwind Program.
The Gemini 4 Argon model increases the maximum token output to 1 million, a significant jump from the previous limit of 64,000 tokens. This expanded capacity enables the model to process lengthy and complex tasks without interruption within a single processing run. Google has set an introductory API price of $2 per million input tokens and $10 per million output tokens. Input tokens stored in cache are discounted by 95%. After this initial phase, prices are expected to rise to $4 and $20, respectively.
According to Google, thousands of its employees are already utilizing Argon for specialized coding, research, and writing tasks. Internal teams have applied the model to optimize data center memory and facilitate large-scale code migrations. One project involved using Argon agents for migrating code from C and C++ to Rust, while another focused on memory profiling across Google’s data centers. Google stated that these optimizations have freed over 300 tebibytes of memory, with additional savings identified through ongoing work.
Enhanced capabilities for demanding professional tasks
Google reported a 77.9% performance score for Gemini 4 Argon on DeepSWE v1.1, a benchmark for long-duration software engineering performance. The company also highlighted its results in finance, legal, and automation benchmarks. The model supports multimodal reasoning, combining coding with enterprise-level workflows. Developed by Google DeepMind as part of the broader Gemini family, Argon’s increased output capacity enables it to better manage extended workflows that involve multiple stages of reasoning and execution.
Cybersecurity is another key area for the model’s initial deployment. Google mentioned that Argon can identify, verify, and patch software vulnerabilities in controlled defensive environments. Wiz is utilizing Argon through its Scan for Good initiative, which aims to identify security vulnerabilities in public infrastructure. The company reported that Argon achieved a score of 68% on CWE-bench v1, a benchmark for vulnerability remediation. Selected cybersecurity defenders are being granted access to the model without standard guardrails to support critical defensive operations.
Initial limited release sets the stage for broader availability of Gemini 4
Google has not yet announced a definitive date for the wider public release of Gemini 4 Argon. The company explained that it is conducting a phased rollout and collecting feedback from early testers. It also participates in a voluntary pre-release model access process with the U.S. government. Future availability will include developers, enterprises, and consumers. The rollout will start with paid API clients and Google AI Ultra subscribers, although no specific launch date has been provided for these groups.
Additionally, Google confirmed it has no plans to release Gemini 3.5 Pro, which was initially scheduled for June. This decision leaves Gemini 4 Argon as the company’s latest flagship model tailored for demanding reasoning and professional tasks. While other Gemini models remain available for different performance and budget needs, Argon distinguishes itself with its larger output, advanced coding features, and specialized cybersecurity capabilities. Currently, access is limited to trusted testers and selected cybersecurity partners involved in defensive security efforts.