Google Drops Gemini 4 "Argon" — 1M Tokens of Reasoning, Still Can't Explain $ETH

Google released Gemini 4 Argon on October 13 through its Gemini app and Search AI Mode, opening its newest reasoning model to Pro and Ultra subscribers alongside developer and enterprise channels. The company positions Argon for coding, legal, financial, cybersecurity, and scientific workloads, with text output and multimodal input support for text, images, video, audio, and PDF documents.

Developers can access Argon via Google AI Studio and the Gemini API, while businesses connect through Google Cloud. Google AI Ultra subscribers can run the model inside Antigravity, the company's agent-based software development environment. Google set introductory pricing at $2 per million input tokens and $10 per million output tokens, with both rates doubling after November 30 when the promotional period ends. The model's documentation lists a 1,048,576-token context window and multiple levels of reasoning effort.

Google reported internal benchmark gains tied to Antigravity usage, including more than 35 percent additional agentic code submissions during testing. The company said engineering teams using Argon reduced code rollbacks by nearly a third and cut session preparation time for its Agent Development Kit by 40 percent. Google attributed the improvements to Argon's reasoning capabilities without releasing the underlying test methodology.

The launch lands in an increasingly competitive market for paid reasoning models, where developers weigh context window size, token pricing, and tool integration alongside raw benchmark scores. With introductory rates scheduled to expire in late November, pricing will likely factor into adoption decisions for studios and cloud customers evaluating sustained workloads. Google has not announced any token-based integrations for $BTC, $ETH, or other digital assets within the Argon platform.