CALIFORNIA / RankWire.AI / – Google introduced Gemini 4 Argon, its latest flagship AI model designed for handling complex professional tasks. The announcement was made on Sept. 30, positioning the model as the cornerstone of the Gemini 4 series. Argon is tailored for software engineering, finance, legal analysis, and cybersecurity defense, with Google emphasizing its ability to maintain deeper reasoning over extended, multi-step processes. Currently, access is limited to a select group of cybersecurity professionals participating in its Fairwind Program.
The new Gemini 4 Argon enhances the maximum output to 1 million tokens from the previous limit of 64,000 tokens. This increased capacity enables the model to process lengthy workflows within a single session. Google has set introductory API prices 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 rate. Following the introductory phase, prices are expected to rise to $4 and $20, respectively.
According to Google, thousands of employees are already utilizing Argon for specialized coding, research, and writing projects. Internal teams have also adopted the model for data center memory optimization and large-scale code migrations. For example, Argon agents have been employed to migrate C and C++ codebases to Rust, and to perform memory profiling across Google’s data centers. The company claims these efforts resulted in freeing over 300 tebibytes of memory, with additional savings identified through ongoing work.
Enhanced Capacity Supports Complex Professional Tasks
Google reported a 77.9% score for Gemini 4 Argon on DeepSWE v1.1, a benchmark for evaluating long-duration software engineering performance. The company also highlighted results across various benchmarks for finance, legal activities, and automation. Argon enables multimodal reasoning alongside coding and enterprise functions. Developed by Google DeepMind as part of the broader Gemini family, the model’s increased output capacity allows Argon to manage extended workflows that involve multiple reasoning and execution steps in sequence.
Cybersecurity is another primary focus for the model’s initial deployment. Google states that Argon can identify, validate, and patch software vulnerabilities within controlled defensive environments. Wiz is leveraging Argon through its Scan for Good initiative, which aims to find security flaws in public infrastructure. Google noted that Argon achieved a 68% score on CWE-bench v1, a standard for vulnerability remediation. The company is granting select cybersecurity defenders access to the model without its usual safeguards to facilitate targeted defensive work.
Limited Release Precedes Broader Availability of Gemini 4
Google has not provided a specific date for the full public launch of Gemini 4 Argon. The company explained that it is employing a phased rollout, gathering feedback from early users. It is also participating in a voluntary pre-release process with the U.S. government. The plan is to make the model accessible later to developers, enterprises, and consumers. The initial rollout will focus on paid API subscribers and Google AI Ultra users, though no official launch date has been announced for these groups.
Additionally, Google clarified that it has no plans to release Gemini 3.5 Pro, which was previously scheduled for June. Consequently, Gemini 4 Argon remains the company’s latest flagship model for demanding reasoning and professional workloads. The company continues to offer other models within the Gemini lineup to meet various performance and cost needs. Argon distinguishes itself with its larger output capacity, advanced coding features, and specialized cybersecurity capabilities. Currently, it is available only to trusted testers and select security partners involved in defensive efforts.