CALIFORNIA / RankWire.AI / – Google has revealed Gemini 4 Argon, its latest flagship artificial intelligence system designed to handle intricate professional tasks. The company introduced this model on Sept. 30 as the core component of its Gemini 4 series. Argon is tailored for use in software engineering, finance, legal analysis, and cybersecurity defenses. Google emphasized that the model can perform more profound reasoning across extended, multi-step procedures. Currently, access is limited to a select group of cybersecurity professionals participating in its Fairwind Program.
The Gemini 4 Argon model significantly increases its maximum output to 1 million tokens from the previous cap of 64,000 tokens, enabling it to process longer, more complex tasks within a single run. The introductory API pricing is set at $2 per million input tokens and $10 per million output tokens, with cached input tokens qualifying for a 95% discount on the input price. Following the initial launch phase, prices are scheduled to double to $4 and $20, respectively.
According to Google, thousands of its employees are already utilizing Argon for specialized coding, research activities, and writing projects. Internal teams have also employed the AI for tasks like optimizing data center memory and executing large-scale code migrations. Notably, one project involved deploying Argon agents for migrating C and C++ codebases to Rust, while another used these agents for memory profiling across Google’s data centers. The company reports that these efforts have freed over 300 tebibytes of memory, with additional savings identified through similar optimizations.
Enhanced Capabilities for Complex Enterprise Tasks
Google shared that Gemini 4 Argon achieved a 77.9% score on DeepSWE v1.1, an assessment of long-term software engineering performance. The company also highlighted its strong results in financial, legal, and automation benchmarks. The model supports multimodal reasoning alongside coding and enterprise-oriented functions. Developed by Google DeepMind as part of the broader Gemini family, Argon’s increased output capacity enables it to manage prolonged workflows that demand multiple consecutive reasoning and execution steps.
Cybersecurity application is another key area in its initial deployment, with Google stating that Argon can identify, verify, and patch software vulnerabilities within controlled defensive environments. Through its Scan for Good initiative, Wiz is utilizing Argon to detect security vulnerabilities in public infrastructure. Google also reported that Argon achieved a 68% score on CWE-bench v1, a benchmark for vulnerability remediation, and is granting select cybersecurity defenders access without the usual safety protocols for approved defense activities.
Limited Deployment Precedes Broader Access to Gemini 4
Google has not set a specific date for the widespread public release of Gemini 4 Argon, but it confirmed a phased rollout is underway, with ongoing feedback collection from early testers. The company is also participating in a voluntary U.S. government process to gain pre-release access to its models. Future availability plans include developers, enterprises, and consumers, with initial deployment targeting paid API users and Google AI Ultra subscribers; however, no fixed launch date has been provided for these groups.
Additionally, Google announced that it has no plans to release Gemini 3.5 Pro, which was previously scheduled for June. This decision leaves Gemini 4 Argon as the newest flagship for demanding reasoning and professional tasks. The company continues to offer other Gemini models tailored to different performance levels and budget considerations. Argon stands out with its larger output limit, advanced coding capabilities, and specialized cybersecurity functions, though current access remains limited to trusted testers and select security partners for the time being.