Google unveils its next artificial intelligence model, i.e. Gemini 4 Argonits availability will initially be limited to a small group of testers. This model aims to provide More advanced and efficient performanceespecially in planning and managing complex tasks of a different nature.
new features
As expected, this is a model designed to support deep inference Long and particularly complex workflowsmanages activities that require high levels of analysis and complex processing capabilities. The areas of application are diverse, from software engineering to the legal and financial sectors to IT security, and the goal is to provide high performance even when dealing with complex professional tasks.
To support its functionality, Google is expanding outgoing token limits. Increased from 64,000 tokens previously to 1 million tokens. This way, the model can even solve advanced problems in one step thanks to the additional available processing space. Various benchmarks also emphasize activities such as programming and the Management of legal tasksas demonstrated by the results obtained in “Harvey’s Lawyer.”
Additionally, due to his visual understanding, Argon can Drill down into charts and act upon the information contained in various documents. However, in terms of cybersecurity, this model can Independently identify and correct software vulnerabilities. Currently, Google has provided it to some network security experts for the purpose of testing its functionality. Therefore, the company will first collect feedback from testers to improve the security system before extending access to developers, enterprises, and users.
Gemini 3.8 Live introduces avatars with faces and voices for more interactive conversations
Availability and cost
Gemini 4 Argon is currently available to limited testers. Expected pricing for API customers is $2 per million incoming tokens For every 1 million tokens withdrawn, there will be a reward of $10. At this time, Google hasn’t specified an actual date for its debut, so we’ll update you as soon as we learn more.