AI
Google's Gemini 4 Argon starts with cyber defenders only
Google released Gemini 4 Argon on Sept. 30, but only to "a set of trusted cyber defenders through our Fairwind Program" at first. Broader access is planned "starting with paid API customers and Google AI Ultra subscribers." So this is the model behind the recent "best coding model" chatter, and most people cannot use it yet.
Google lists 77.9% on DeepSWE v1.1 (a coding test) and 68% on CWE-bench v1 (a security-bug test, tied for first), plus 1 million tokens of output, up from 64K. Introductory pricing is $2 per million input tokens and $10 per million output, rising to $4 and $20 later. Google also says Argon agents replaced 32K lines of SIMD code (a way of doing many calculations at once) in a Rust port, making video decoding "2.7x faster" than the earlier Rust port, closer to optimized C++. These are Google's own numbers.
On your desk: the crown for best model keeps rotating, so avoid hard-wiring one model into your n8n flows and Skool course demos. Keep the model name in one setting you can change, and teach students the same habit.
Read the original at Google blog: Gemini 4 Argon →