Google announces Gemini 4 Argon AI model, but you can't use it yet

Google announced its new frontier artificial intelligence model, Gemini 4 Argon, representing the company’s first frontier model since February. Although Google claims the model offers industry-leading performance in knowledge work, cybersecurity, and coding, the public cannot use it yet because it is currently undergoing limited testing with trusted testers.

Internal testing has already yielded notable results, with Argon utilizing fleet-wide telemetry data to save 300 TiB of memory across data centers and assisting in migrating C/C++ codebases to Rust across Google. Furthermore, Argon achieved a 77.9 percent score on the DeepSWE v1.1 software engineering benchmark and secured an industry-leading score on the Vals Index test.

Google is rolling out the model starting with a small group of trusted testers, including partners in the Fairwind Program focusing on cyber defense capabilities. Wiz is already utilizing Argon to uncover critical vulnerabilities. Gemini 4 Argon supports an output limit of 1 million tokens and features built-in systems to monitor its chain-of-thought.

Google confirmed that general availability will eventually begin with paid API users and Google AI Ultra subscribers, though no specific timeline was provided.