Google announces Gemini 4 Argon with 1M token limit and frontier coding capabilities

Google has announced Gemini 4 Argon, a new frontier model designed for complex, long-horizon workflows in software engineering, enterprise knowledge work, and cybersecurity defense. The model is currently rolling out to trusted cyber defenders through the Fairwind Program and is undergoing pre-release access with the U.S. government before a broader launch for developers and enterprises.

Key Capabilities and Benchmarks

Gemini 4 Argon features an industry-leading 1 million token output limit, up from the previous 64K, enabling deep reasoning across extended tasks. It sets a new state of the art on DeepSWE v1.1 with a score of 77.9% for real-world software engineering and ranks #1 on AutomationBench with a score of 51.3%. In enterprise domains, it leads the Vals Index and achieves state-of-the-art performance on LVBench (91.7%) for long video understanding.

Internal Impact and Cybersecurity

Internally, Argon agents have optimized quantum algorithmic subroutines by beating baselines by 40% and identified memory optimizations freeing over 300 TiB of memory across Google’s data centers. For cybersecurity, the model autonomously finds and patches vulnerabilities, scoring 68% on CWE-bench v1. It is being used by Wiz in its Scan for Good initiative, where it uncovered a critical vulnerability in healthcare software that previous models missed.

Pricing and Safety

Argon will launch at an introductory price of $2 per million input tokens and $10 per million output tokens, with cached inputs discounted by 95%. Google is implementing rigorous safeguards, including monitoring internal activations for misuse and hardening sandboxed environments, before making the model available to paid API customers and Google AI Ultra subscribers.