DeepMind details Argon's 1M token output for solving long multi-step problems in one go

GoogleDeepMind · x · 2026-10-01

Google DeepMind elaborated on Gemini 4 Argon's headline capability: a 1M token output limit that lets the model tackle long, multi-step problems in a single pass. Feedback from early testers will help strengthen systems before broader rollout to developers, enterprises, and consumers soon.

Related event: DeepMind Opens Gemini 4 Argon via Fairwind with 1M-Token Output(2 posts)→

Original post →

More from Models

Models channel →