Google has introduced Gemini 4 Argon, a new AI model designed for complex, long-running tasks across software development, enterprise work and cybersecurity.
The biggest change is its much larger output capacity. Gemini 4 Argon can generate up to 1 million output tokens, compared with the previous 64,000-token limit. Google says the expanded limit is designed to help the model work through lengthy, multi-step problems without having to stop and split the task into smaller parts.
Also read: How to Use Google Find Hub to Remember Where You Put Your Things
Gemini 4 Argon Availability
Gemini 4 Argon is not being released to everyone immediately. Google is first making it available to selected cybersecurity professionals through its Fairwind Program, allowing trusted testers to evaluate its capabilities in real-world security work.
After this initial testing phase, Google plans to expand access to paid API customers and Google AI Ultra subscribers, followed by developers, enterprises and consumers. Google says it is using feedback from early testers to improve safeguards before a wider rollout.
The introductory API pricing is set at $2 per million input tokens and $10 per million output tokens. Google says the pricing will later increase to $4 per million input tokens and $20 per million output tokens after the introductory period.
Gemini 4 Argon Features and Capabilities
The model is built for tasks that require extended reasoning and multiple steps. Google says Argon has achieved a 77.9% score on DeepSWE v1.1 for software engineering, 91.7% on LVBench for long-video understanding and 51.3% on AutomationBench for business automation.
Google is also already using Argon internally. Its applications include debugging, codebase migrations, algorithm development and research. In one example, Argon helped improve a Rust version of Google’s libgav1 video decoder. Google says the resulting decoder runs 2.7 times faster than the existing Rust port while producing the same video output.
Focus on Cybersecurity
Cybersecurity is another major part of Gemini 4 Argon’s launch. Google says the model can identify, validate and patch software vulnerabilities, with a 68% score on CWE-bench v1, tying for the top result on that benchmark.
Because of these capabilities, Google is taking a cautious approach to the rollout. The company is testing safeguards against cyber misuse, indirect prompt injection and other unwanted model behaviour before making Argon broadly available.
Also read: How to Use Google’s New Android Features From the September 2026 Update
Final Thoughts
Gemini 4 Argon is aimed at users who need AI to handle much longer and more complicated workflows. Its 1-million-token output limit, coding capabilities and cybersecurity focus make it a notable addition to Google’s Gemini lineup, although most users will have to wait for the wider rollout.