Google has announced Gemini 4 Argon, its new frontier model built for deep reasoning across long-horizon workflows. It is rolling out first to trusted cyber defenders through the Fairwind Program while Google gathers feedback and strengthens safeguards. Broader access for developers, enterprises, and consumers will follow as soon as possible, starting with paid API customers and Google AI Ultra subscribers. Argon expands the output token limit to an industry-leading 1 million tokens, up from 64,000. Google reports state-of-the-art results on DeepSWE v1.1 at 77.9% for real-world software engineering, leading scores on the Vals Index for finance, legal, and coding work, and 91.7% on LVBench for long-video understanding. The model is already used internally for quantum optimization, large-scale C/C++ to Rust migrations, and memory efficiency gains across data centers.
Google has announced Gemini 4 Argon, its new frontier model built for deep reasoning across long-horizon workflows. It is rolling out first to trusted cyber defenders through the Fairwind Program while Google gathers feedback and strengthens safeguards. Broader access for developers, enterprises, and consumers will follow as soon as possible, starting with paid API customers and Google AI Ultra subscribers. Argon expands the output token limit to an industry-leading 1 million tokens, up from 64,000. Google reports state-of-the-art results on DeepSWE v1.1 at 77.9% for real-world software engineering, leading scores on the Vals Index for finance, legal, and coding work, and 91.7% on LVBench for long-video understanding. The model is already used internally for quantum optimization, large-scale C/C++ to Rust migrations, and memory efficiency gains across data centers.