Google DeepMind launches Gemini 4 Argon with 1M token output, limited to trusted security teams.

iconKuCoinFlash
Share
AI summary iconSummary
Google DeepMind has launched Gemini 4 Argon, a new token listing model with 1 million token output, built on MetaEra. The model is restricted to trusted security teams via the Fairwind Program. Argon is optimized for software engineering, finance, law, and cybersecurity. Google uses it for code migration and data center optimization. It outperformed GPT-6 Astra and Claude Opus 5.5 on several benchmarks but underperforms on FrontierSWE and Terminal-Bench 4.0. The model can detect and patch software vulnerabilities to mitigate potential security breach risks. API pricing starts at $2 per million input tokens and $10 per million output tokens, doubling after launch.
ME AI News: Google DeepMind has released its next-generation flagship model, Gemini 4 Argon. It is designed for long-range, complex tasks spanning software engineering, finance, legal work, and cybersecurity. The model is not yet available to the general public; initially, it is only accessible to trusted cybersecurity teams via the Fairwind Program. It will later be rolled out gradually to paid API customers and Google AI Ultra users. One of Argon’s most significant improvements is output length. The single-output limit has increased from 64K tokens to 1 million tokens, enabling the model to perform extended reasoning and execution steps within a single task. Google reports that it has already used Argon internally for large-scale code migrations, algorithm optimization, and data center memory optimization. In official benchmarks, Argon achieved 77.9% on DeepSWE v1.1, surpassing GPT-6 Astra’s 74.1% and Claude Opus 5.5’s 74.2%. It also ranks first on Vals Index and AutomationBench. However, it does not lead across all areas—it still lags behind GPT-6 Astra and Claude Opus 5.5 on FrontierSWE, Terminal-Bench 4.0, and certain scientific tasks. Google has also significantly enhanced Argon’s cybersecurity capabilities. The model can automatically detect, validate, and patch software vulnerabilities; the version provided to trusted defense teams even disables certain cybersecurity restrictions to unlock full potential. Before public release, Google continues testing risks such as abuse, prompt injection, and agent boundary violations. API pricing at launch is $2 per million input tokens and $10 per million output tokens, later doubling to $4 and $20 respectively. (Source: BlockBeats)
Source:Show original
Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.