Gemini 4 Argon Targets Long-Horizon Work
Gemini 4 Argon is Google’s new frontier model for professional tasks that require multi-step reasoning, sustained execution, and understanding across documents. Google says it is already used internally for software engineering, research, and writing, while initial external access is going to trusted cyber defenders through the Fairwind Program.
Gemini 4 Argon Expands Enterprise Workflows
The model’s output limit rises from 64,000 to one million tokens. Google positions it for large codebase migrations, financial research, legal drafting, chart and long-video analysis, and other enterprise workflows. The company reports leading results on selected internal or partner evaluations, including 77.9% on DeepSWE v1.1 and 51.3% on AutomationBench. Such scores describe benchmark performance, not guaranteed gains in every real-world workflow.
Cyber Defense and Release Constraints
For cybersecurity, Google says Argon can find, validate, and patch software vulnerabilities, tying for first on CWE-bench v1 with a 68% score. Wiz is also testing it through a public-interest program. Broad release has not happened: Google is continuing red-team work, misuse safeguards, and prompt-injection defenses. The central open question is whether such long autonomous task chains can remain verifiable, controllable, and safe in complex environments. Google lists introductory pricing of $2 per million input tokens and $10 per million output tokens, with a separate discount for cached inputs.