Google Unveils Gemini 4 AI Model, Argon, in Limited Availability

Google has finally released its Gemini 4 AI model, known as Argon, but only to a select few organizations through its Fairwind Program. The limited launch is due to the US government's voluntary process for granting early access to models to test and improve their safety guardrails before making them widely available.

The release of Argon comes after months of delay and the postponement of Gemini 3.5 Pro, which was initially promised for June. Google DeepMind head Koray Kavukcuoglu had previously stated that Gemini 4 should be released 'much earlier' than the end of this year.

Argon boasts an increased token capacity, with introductory pricing set at $2 per million input tokens and $10 per million output tokens. The model's output limit has been raised from 64,000 tokens to 1 million, allowing for more complex tasks within a single trajectory.