Gemini 4 Argon: our next era of frontier intelligence
DeepMind unveiled Gemini 4 Argon, a phased frontier model aimed at coding, knowledge work, and cyber defense.
Google DeepMind announced Gemini 4 Argon, offering early access to trusted defenders via the Fairwind Program during a phased U.S. government pre-release process. The model is priced at $2 per million input tokens and $10 per million output tokens, with a 1 million token output limit. Reported scores include 77.9 percent on DeepSWE v1.1, 51.3 percent on AutomationBench, and 91.7 percent on LVBench. Internal agents freed over 300 TiB of memory and are rewriting large C/C++ codebases in Rust, including the Fuchsia Zircon kernel.
- DeepMind is phasing access while engaging the U.S. voluntary pre-release process.
- Output limit increases from 64,000 tokens to 1 million tokens.
- Argon leads cited coding, automation, video, and economic-work benchmarks.
- Internal runs freed over 300 TiB, with up to 1 PiB projected.
- Rust migrations include re2, libgav1, and 800,000-plus Zircon kernel lines.
- Vendors
- Google DeepMind
- Products
- Gemini 4 Argon
- Organizations
- GoogleGoogle DeepMindWiz
- Countries
- United States
- AI models
- Gemini 4 Argon
From the original article
6 min read at deepmind.google
Gemini 4 Argon delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense. This content is generated by Google AI. Generative AI is experimental Today, we’re announcing our new frontier model, Gemini 4 Argon, which is rolling out to a set of trusted cyber defenders through our Fairwind Program .
deepmind.google
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.