PH
Gemini 4 Argon is here—Google's new frontier model is built for the hard stuffGoogle has unveiled Gemini 4 Argon, a model designed for long-running reasoning and real-world work: software engineering, professional research and defensive cybersecurity. The headline feature? Google says it can generate up to one million output tokens in a single run—up from the previous 64K output limit. That's an output limit, not a claim about the input context window.
What stands out- Coding and complex workflows: Google says Argon can handle extended software-engineering tasks, including debugging and large codebase migrations.
- Cyber defense: It is being tested with trusted defenders for finding, validating and patching vulnerabilities.
- Published benchmark results: Google reports 77.9% on DeepSWE v1.1, 51.3% on Zapier's AutomationBench and 91.7% on LVBench. These are reported evaluation results, not a promise that every task will perform the same way.
Can you try it yet?Not broadly. Argon is initially rolling out to trusted cyber defenders through Google's Fairwind Program. Google plans to expand access to developers, enterprises and consumers, starting with ρáíd API customers and Google AI Ultra subscribers. No firm general-release date was announced.
Announced API pricingIntroductory pricing is $2 per million input tokens and $10 per million output tokens, with cached input priced at 95% off. After the introductory period, Google says the rates will be $4 input / $20 output per million tokens. This is not a free-access announcement; check actual availability and billing when access opens.
Bottom line: Gemini 4 Argon looks especially interesting for coding agents and long, multi-step professional tasks. The capabilities are exciting, but the rollout is deliberately limited for now. What use case would you test first once it becomes available?
Official source: You do not have permission to view the full content of this post. Log in or register now.