Anthropic published Claude 3.7 Sonnet, a new argumentation model

Anthropic has announced the publication of Claude 3.7 Sonnet, a new hybrid argumentation -KI model, together with a special tool for developers named Claude Code. Claude 3.7 Sonett has so far been described as the most advanced model of the company and introduces a unique approach by combining fast reaction functions with expanded, self -reflecting thinking. With this dual mode functionality, users can choose between quick answers or a more detailed, gradual problem solution. The model is available in all anthropic plans – free, professional, team and enterprise – as well as on platforms such as Amazon DOBATION and the Vertex AI from Google Cloud. However, the extended mode of thought is excluded from the free level.

The most important improvements in Claude 3.7-Sonnet include an improved performance in coding and front-end-web development tasks, with which it is particularly effective for real applications such as debugging, planning code changes and the treatment of complex workflows. The model also introduces fine-grained control over its “thinking budget” through the API integration, so that users can set limits for the computing effort that the model should spend on a task.

Accompany this version Claude Code, a command line tool that aims at Workflows of software development. Claude Code is currently available as a limited research preview and enables developers to automate tasks such as reading and editing code, executing tests and managing Github repositories directly from your terminal. Early tests have shown that the tool can significantly reduce the development time by doing tasks that normally take over 45 minutes in just one pass.

Anthropic emphasized security and reliability in this publication. The system card for Claude 3.7 -Sonnet outlines measures to mitigate risks such as fast injection attacks and at the same time to improve the ability of the model between harmful and benign inquiries. Compared to its predecessor, it reduces unnecessary rejections by 45%.

This publication reflects Anthropic's wider goal of creating AI systems that are able to improve human skills in different areas. By integrating reasoning into a single model and the concentration on practical applications and not on theoretical benchmarks, Claude 3.7 Bonnet is aimed at determining a new standard for large-scale models in both business and development contexts. Feedback from Claude 3.7 Sonnet and Claude Code users play a crucial role in the design of future updates and improvements in these tools.

source