Gemini 4 Argon

Gemini 4 Argon is a frontier model released to trusted cyber defenders via Google’s Fairwind Program. It supports a 1 million-token output limit and charges $2 per million input tokens and $10 per million output tokens. The model accelerates quantum algorithmic optimization by 40%, frees 300 TiB of memory across data centers, and migrates up to 800 k lines of C/C++ to Rust with 2.7× speed gains for libgav1. Gemini 4 Argon scores 77.9% on DeepSWE v1.1 and leads the Vals Index for finance, coding, and legal tasks.