Gemini 4 Argon Just Went Live for Everyone — Pricing, Tasks Mode, and What's New

Google's most powerful AI model is out of early access. Gemini 4 Argon, first announced on September 30 and initially limited to trusted cybersecurity defenders through Google's Fairwind Program, launched globally on October 13, 2026 — for developers, enterprises, and consumers alike.

The model is now available across Google's whole stack: the Gemini API, AI Studio, Google Antigravity, the Gemini app, and AI Mode in Search. On the consumer side, Google AI Pro and Ultra subscribers get Argon inside the Gemini app, where Google says it delivers "frontier reasoning" and major improvements in coding.

Pricing is the headline for developers. Introductory API rates are $2 per million input tokens and $10 per million output tokens, with cached input discounted 95%. That intro window runs through November 30, 2026 — after which standard pricing kicks in at $4 per million input and $20 per million output. For a frontier-class reasoning model, that's aggressive.

Under the hood, Argon gets a unified 1,048,576-token context window shared by input and output. The default synchronous output is capped at 65,536 tokens, but a continuation protocol lets it keep generating within the million-token budget. It accepts text, images, video, audio, and PDFs as input — though output is text only: no image generation, no audio generation, and no Live API support, per Google's developer docs. Google reports Argon ranks first on the Vals Index, which measures model performance across finance, coding, legal, and tax work — framed as Google's claim, as with all vendor benchmarks.

The most interesting part isn't the specs, though — it's the behavior. Google says Argon answers more directly with less hedging, repeats itself less, and shares progress updates more frequently during long tasks. And in the Gemini app, a new toggle switches from chat to tasks mode: Gemini can work across your Google apps even when you're not at your computer. Google also says Gemini now connects to more than 40 apps — including Instacart, OpenTable, Adobe, and Shopify — for workflows like ordering groceries, booking restaurant tables, and business design tasks.

Why it matters: the message is unmistakable. Google isn't really selling a chatbot anymore — it's selling an agent that keeps working when you walk away. The coding improvements plus Antigravity's agentic coding baked into the macOS app point the same way. If the launch pricing sticks around past November, the cost of running a genuinely capable agent stack is about to drop hard.

My take: the model was impressive on paper two weeks ago; what makes today's launch interesting is the delivery — consumer availability, clear pricing, and tasks mode. The real test will be whether tasks mode actually completes multi-app jobs unsupervised. That's the demo I'll be watching for.