Google's newest AI model is here, but you can't try it yet. Thousands of Googlers are already using it internally. Google says Argon agents found memory optimizations that will free up more than 300 TiB across its data centers once rolled out. The model's output limit is also now 1 million tokens, up from 64,000. Google says Gemini 4 Argon is the leading model on the Vals Index, which measures economic impact across finance, coding, legal and tax work. Each sector is weighted by its share of US GDP. Google also says it leads on Vals Finance Agent v2, a test of multi-step financial research. Those results come from Google, so the real test is how Argon performs once outsiders get their hands on it. Read more: