Every so often, an AI announcement actually matters — not because of the name, but because of what it lets you do next. Google just announced Gemini 4 Argon, its newest frontier model. Two things make it stand out. First, deep reasoning: Argon is designed to think through hard, multi-step problems more carefully before answering. Second, output size: it can produce up to 1M tokens in a single response — in everyday terms, it can hold its train of thought across very long answers, documents, and projects instead of losing the thread partway through.

Who gets it first

Argon is rolling out first to developers and trusted cyber defenders — the people who stress-test AI for security work. For builders, Google set introductory API pricing at $2 / $10 per 1M tokens. That pricing matters to you more than it might seem: when the engine behind an app gets cheaper and stronger, the apps built on it tend to get better faster — and often cheaper to run.

What to watch as a regular user

To be clear, this part is a watchlist, not a promise — Google hasn't announced consumer dates yet. But frontier models like Argon usually preview what everyday Gemini becomes next:

Who should try it

Developers can start building with Argon now through the API. For everyone else, the action item is simple: keep an eye on Gemini. If you already use it for writing, research, or planning, the next round of upgrades is likely to feel like a real step up — and we'll walk you through each one the day it lands.

New model, same rule here: what can you do with it? That's the only benchmark that counts.

Watch the 30-second version The Gemini 4 Argon announcement, in one quick drop Back to The Tech Drop home