Gemini 4 Argon is a 📝Google product and the first model in the Gemini 4 generation of 📝Gemini, built by 📝DeepMind and announced on October 13, 2026 as Google's new 📝frontier model. It is designed for deep, multi-step work in coding, finance, law, cybersecurity, and science, and offers a 1-million-token combined input and output limit through the Gemini API. It follows a year in which Google shipped Gemini 3.x Flash models every few weeks while its top Pro model, Gemini 3.1 Pro, stayed in preview; Google reports that Argon ranks first on the Vals Index and scores 87% on the ERQA embodied-reasoning benchmark, with the 📝rest of the Gemini 4 family still to come. Argon is available globally to developers through the Gemini API, AI Studio, and Android Studio, to enterprises through Gemini for business and Google Antigravity, and to Google AI Pro and Ultra subscribers in the Gemini app and AI Mode in Search.
Key Features
- Unified 1M-token limit — Input and output share a single 1-million-token budget in the Gemini API, and a new continuation protocol lets very long generations resume where they stopped.
- Agentic coding — Google reports over 35% more agentic code submissions and nearly a third fewer rollbacks in Antigravity, plus a Chrome refactor finished in weeks rather than months.
- Science and research — Google cites faster quantum chemistry and quantum error decoding, new counterexamples in fusion research, and five enzyme designs confirmed active in the lab.
- Robotics and embodied reasoning — A Video-to-Sim-to-Real pipeline teaches robots from video; in one internal test, an Apptronik Apollo humanoid sank seven cornhole shots in a row.
- Layered safety stack — Tested under Google's Frontier Safety Framework, with jailbreak defenses, a real-time misalignment monitor, and indirect prompt-injection success cut to 0.7%.
Getting Started
- Open Google AI Studio on a paid Gemini API tier and select Gemini 4 Argon as your model.
- Expect $2 input and $10 output per million tokens until November 30, then $4 and $20.
- Cache repeated inputs where you can, since cached input tokens are billed at 95% off.
- Or use it in the Gemini app and AI Mode in Search with a Google AI Pro or Ultra plan.