Google Unveils Gemini 4 Argon, its Most Powerful AI Model Yet

Google has unveiled its most powerful AI model yet, called Gemini 4 Argon. This model is designed for long and complex tasks such as software engineering, legal work, and cybersecurity defense. It tops 13 out of 18 benchmarks against OpenAI and Anthropic models.

Gemini 4 Argon scored 77.9% on DeepSWE v1.1, surpassing Claude Opus 5.5 at 74.2% and GPT-6 Astra at 74.1%. It also leads the Vals Index, which weights finance, coding, legal and tax work by their share of US GDP, with 68.9%.

However, rivals still hold some ground in certain areas, with GPT-6 Astra staying ahead on FrontierSWE v2 (65.5% to 55%) and OSWorld-2.0. Claude Opus 5.5 wins Terminal-bench 4.0 and PostTrainBench.

Google is already using Argon to rewrite code and free up data centre memory, with thousands of Googlers having used it internally for weeks. The company estimates that this will save around 500 TiB to 1 PiB of memory.