CALIFORNIA / RankWire.AI / – Google announced on September 30 the launch of Gemini 4 Argon, its new flagship AI model designed for sophisticated professional tasks. The company positioned Argon as the lead model within the Gemini 4 series, emphasizing its capabilities in complex software development, financial analysis, legal research, and cybersecurity operations. This advanced model also excels at managing extended sequences of reasoning and execution. Access to Argon remains limited, with select cybersecurity experts participating through the Fairwind Program.
The maximum token output for Argon has been increased to 1 million, a significant rise from the previous cap of 64,000 tokens. This enhancement enables the model to undertake longer, more intricate tasks without needing to split work across multiple sessions. Initial API pricing begins at $2 per million input tokens, while output tokens are billed at $10 per million during the same period. Inputs cached for reuse are discounted by 95%. Future pricing adjustments will see rates rise to $4 for input tokens and $20 for output tokens.
Currently, thousands of employees within Google are utilizing Argon for coding, research, and content creation tasks. The internal teams have already tested the model on projects involving data center optimization and large-scale software migrations. One such project employed Argon agents to facilitate the migration of C and C++ codebases to Rust, while another focused on memory profiling across data centers. These efforts resulted in freeing more than 300 tebibytes of memory, with ongoing analysis revealing further potential savings across the same systems.
Enhanced Long-Form Technical Support in Argon
Google reported a 77.9% performance score for Argon on DeepSWE v1.1, a benchmark measuring extended software engineering tasks. The company also shared results in areas such as finance, legal research, automation, and multimodal applications. Argon was developed by Google DeepMind as part of the broader Gemini model family, which integrates coding tools with long-context reasoning and multimodal capabilities. Its expanded output capacity makes it suitable for workflows requiring multiple interconnected steps before completion.
Cybersecurity remains a core focus in the initial deployment. Argon can detect, verify, and remediate software vulnerabilities within authorized security environments. Through its Scan for Good initiative, Wiz is leveraging the model to identify security flaws in public infrastructure. Google also reported a 68% score on CWE-bench v1, a benchmark dedicated to vulnerability remediation. Furthermore, selected cybersecurity teams can use Argon outside of standard guardrails for approved defensive security tasks.
Limited Public Access During Gradual Rollout
There is no fixed date yet for broad public availability of Gemini 4 Argon. Google is adopting a phased approach, collecting feedback from early users, and participating in a voluntary U.S. government process that grants pre-release access to advanced AI models. Eventually, the model will be accessible to developers, enterprise clients, and consumers. Priority access is expected for paid API subscribers and Google AI Ultra members, although no official launch date has been set.
The company also clarified that it does not plan to release Gemini 3.5 Pro, which had been anticipated prior to the Gemini 4 release. Instead, Argon now represents the latest flagship model aimed at demanding reasoning and professional workloads. Other Gemini models will continue to be available for users with different performance and pricing requirements. For now, Gemini 4 Argon is primarily available to trusted testers, cybersecurity partners, and select early-access programs, with wider availability still pending.