CALIFORNIA / RankWire.AI / – Google has unveiled Gemini 4 Argon, its latest top-tier artificial intelligence designed for complex professional tasks. The company announced Argon on Sept. 30 as the flagship model in the Gemini 4 series. It is capable of supporting advanced software creation, financial analysis, legal research, and cybersecurity efforts. The model also manages extended reasoning and execution sequences. Currently, access is limited, with select cybersecurity specialists testing Argon through the Fairwind Program.
Argon significantly raises the maximum token output to 1 million, a notable jump from the former limit of 64,000 tokens. This increase enables the model to handle longer projects without splitting work into multiple sessions. Initial API pricing starts at $2 per million input tokens. Output tokens are priced at $10 per million during the same period. Inputs stored in cache benefit from a 95% discount. Later, prices are expected to rise to $4 for input tokens and $20 for output tokens.
Many employees within the company are already using Argon for coding, research, and writing tasks. Internal teams have employed the model for data center optimization and large-scale software migration projects. One example involved using Argon agents to support migrations from C and C++ to Rust. Another project focused on memory profiling across data centers. These efforts resulted in freeing more than 300 tebibytes of memory. The teams also discovered further savings through ongoing system analysis.
Extended technical tasks now better supported by Argon
Google reported a 77.9% performance score for Argon on DeepSWE v1.1, a benchmark for long-term software engineering tasks. The company also shared results in finance, legal research, automation, and multimodal tasks. Argon was developed by Google DeepMind as part of the broader Gemini model family. It combines coding capabilities with long-context reasoning and multimodal functions. Its expanded output capacity helps with projects requiring many interconnected steps to complete.
Cybersecurity remains a key focus during the initial rollout. Argon can detect, verify, and fix software vulnerabilities in approved security environments. Wiz is using the model through its Scan for Good initiative, which aims to find security weaknesses in public infrastructure. Google reported a 68% score on CWE-bench v1, a benchmark for vulnerability remediation. Selected security teams can also utilize Argon without standard cyber safeguards for approved defensive security work.
Limited public access during phased deployment
Google has not announced a specific date for widespread public access to Gemini 4 Argon. The release is happening in phases, with feedback being collected from early users. The company is also participating in a voluntary U.S. government process that allows pre-release access to advanced AI models. Later, the model will be available to developers, enterprise clients, and consumers. Priority access will likely go to paid API users and Google AI Ultra subscribers, though no official launch date has been set.
Google confirmed it will not release Gemini 3.5 Pro. This model was expected before Gemini 4. Argon now represents the top-tier flagship for demanding reasoning and professional uses. Other Gemini models are still available for different performance and price points. For now, Gemini 4 Argon remains limited to trusted testers, cybersecurity partners, and early-access programs while broader release plans are still underway.