Google DeepMind has introduced Gemini 4 Argon, a new model designed for complex tasks in software development, enterprise knowledge management, and cybersecurity. The model is capable of generating up to 1 million output tokens in a single response, allowing for extensive code refactoring and long-form report generation without the need for multiple interaction turns.
The company has implemented a phased release strategy, participating in the U.S. government’s voluntary pre-release model access program to collect feedback and refine safety guardrails. Currently, the model is available to select partners, including Wiz, which utilized Argon to identify a critical vulnerability in global healthcare software that previous models had failed to detect.
Pricing for Argon starts at an introductory rate of $2 per 1 million input tokens and $10 per 1 million output tokens. Cached input tokens are priced at $0.10 per 1 million. Following the introductory period, rates will increase to $4 for input and $20 for output. The model is also designed to autonomously detect and patch software vulnerabilities.
In performance evaluations, Google reports that Argon outperformed competitors such as GPT-6 Astra, Claude Opus 5.5, and Claude Fable 5.1 on 12 out of 18 benchmarks, while tying for first place on one. Artificial Analysis noted that Argon matches the performance of GPT-6 Astra on its Intelligence Index while operating at 60 percent of the cost per task.
