DeepMind Launches Gemini 4 Argon with Industry-First 1M Output Tokens
Google DeepMind has officially launched its new frontier model, Gemini 4 Argon, achieving state-of-the-art results in 13 out of 19 major benchmarks. The model's standout feature is the experimental 'Long Decode Continuation' technique, which extends output capacity to 1 million tokens—an industry first. Currently, access is restricted to trusted testers in cybersecurity via the Fairwind Program, with broader availability promised soon.
Related Entities
Full Report
GDM last shipped a larger-than-Flash model in February (3.1 Pro), and after successive incremental 3.x Flash versions and the big GDM management shakeuplast month, the largest question for GDM was when they would catch up to peers who have in the meantime launched Fable and Astra class models.
Well, Argon’s here, with VERY respectable benchmarks (SOTA in 13 of 19 credible benchmarks)… but only accessible in limited cybersecurity preview, though access is promised “as soon as possible”:
[Google DeepMind@GoogleDeepMindIntroducing Gemini 4 Argon – our new frontier model. It’s built for complex workflows across coding, enterprise knowledge work, and cybersecurity defense – rolling out today to a set of trusted testers through our Fairwind Program.

8 PM · Sep 30, 2026 · 3.42M Views
](https://x.com/GoogleDeepMind/status/2105388084154056939)
We like the experimental Long Decode Continuation, which increases output tokens up to 1M as an industry first.