Google Unveils Gemini 4 Argon with 1M Output Tokens and 77.9% Coding Benchmark Lead

Google has announced Gemini 4 Argon, its new frontier-level AI model designed for complex, long-horizon workflows across software engineering, enterprise knowledge work, and cybersecurity defense. The model achieves a state-of-the-art 77.9% pass rate on the DeepSWE v1.1 coding benchmark, outperforming Anthropic's Claude Opus 5.5 at 74.2% and OpenAI's GPT-6 Astra at 74.1%.