Anthropic company introduced a new and powerful language model

Anthropic, one of the leading companies in the field of artificial intelligence technologies, has officially unveiled its new Claude Sonnet fifth-generation language model, as reported by Zamin.uz.

This update represents the most balanced member of the Claude family, bringing its performance in complex tasks significantly closer to that of the flagship Claude Opus. The key feature of the new model is that it maintains high accuracy while being offered to users at a much lower cost.

According to developers, the new model is primarily designed for autonomous agent systems—scenarios where AI can perform specific tasks sequentially and independently. The model does not just work with text; it is capable of independently completing complex multi-step tasks using a web browser, command-line tools, and various software interfaces.

Experts note that the new model has significantly surpassed previous generations in planning and working with external tools. In terms of capabilities and technical specifications, it has shown substantial growth in four main areas:

logical reasoning, programming, working with external tools, and completing complex tasks. Internal tests conducted by the company show that in some cases, the new model achieves results comparable to much more expensive high-end models.

In particular, the model’s performance in specialized tests measuring effectiveness in real computer environments has been rated very highly. According to early user feedback, this model is much more stable when completing long-running processes.

Where earlier versions sometimes left complex tasks incomplete, the new generation can independently verify its results and correct errors without user prompting. This significantly streamlines workflows for developers and data analysts, saving time.

Security remains a top priority for developers. The new model is notably more resistant to harmful requests and attacks aimed at deceiving the system.

Additionally, instances of the model presenting false information as fact—commonly known as AI hallucinations—have been significantly reduced. It should be emphasized that the model has not been specifically trained to counter cyberattacks, and its capabilities in this area are limited.

The new model is currently available to all users, including those on free and paid plans. It is also offered through specialized software platforms and interfaces.

The company has introduced affordable pricing, making modern technologies accessible not only to large enterprises but also to small and medium-sized businesses. IT specialists and developers in Uzbekistan can use this update to automate their workflows and improve efficiency.