Anthropic releases Sonnet 4.6: fewer errors, available to everyone

Anthropic releases Sonnet 4.6: fewer errors, available to everyone

Anthropic, one of the leading players in the AI market, has unveiled an updated version of its flagship mid-tier language model, Sonnet 4.6. The release marks an important step in generative AI, as the developers concentrated on aspects that directly affect user experience and practical use. The main emphasis is on significantly better code generation, higher accuracy on tasks that require long-context processing, and more reliable execution of user instructions. Particular attention went to minimizing so-called "hallucinations", where an AI generates false or fabricated information, one of the main problems of modern language models. In addition, the Computer Use feature, designed to automate user actions in digital interfaces, has been substantially reworked.

In recent years, the rapid growth of large language models (LLMs) has produced many tools that can handle a wide range of tasks, from writing and translation to programming and data analysis. Despite impressive progress, however, models often run into limits on accuracy, reliability and understanding of complex instructions. Anthropic, founded by former OpenAI staff, has said since its founding that it aims to build safer and more ethical AI systems. Sonnet, which sits between fast but less powerful models and slower but more capable heavyweights, has always been positioned as a balanced option. The Sonnet 4.6 update continues that strategy, offering improved performance without a significant loss of speed.

The key improvements in Sonnet 4.6 deserve a closer look. In code generation, the model now shows higher accuracy and efficiency, making it a valuable tool for developers.

Its ability to process and analyze information from very long texts, up to a million tokens according to the developers, opens new possibilities for working with large documents, books and big datasets. This helps the model grasp nuance and context, which is critical for complex reasoning and decision-making. Reduced hallucinations are arguably the most important improvement for the broad user base.

Fewer false or incorrect answers build trust in the model and make it better suited to critical applications. The Computer Use feature, which lets the AI interact with a computer and perform actions, also received significant improvements, which promise smoother and more intuitive workflow automation.

One of the most significant aspects of the Sonnet 4.6 release is its availability. The model is now the default choice in the Claude chatbot, Anthropic's flagship product. Notably, Sonnet 4.6 is available both to free users and to subscribers of the paid Claude Pro plan. This makes advanced AI technology accessible to a wider audience and helps it spread into everyday life and professional work. It also strengthens Anthropic's market position by offering a competitive product that combines intelligence, speed and reliability, while remaining available to everyone.

In conclusion, the release of Sonnet 4.6 is a significant achievement for Anthropic and an important step forward for the AI industry. Improvements in code generation, long-context work, instruction-following accuracy and, above all, fewer hallucinations make this one of the most promising models on the market. Open access to Sonnet 4.6 through the Claude chatbot, for both free and paying users, democratizes access to advanced AI technology. The update confirms Anthropic's aim to build powerful, reliable and accessible AI solutions that can benefit society.