Anthropic Unveils Claude Sonnet 4.5 with Enhanced Coding Features

Anthropic has launched Claude Sonnet 4.5, a new AI model that boasts enhanced coding capabilities and a suite of new features designed to streamline development processes. In a company announcement, Anthropic described Claude Sonnet 4.5 as "the best coding model in the world," highlighting its ability to build complex agents and improve computer usage.

One of the standout features of Claude Sonnet 4.5 is the introduction of "checkpoints," which allow developers to save their progress and revert to previous states if needed. This feature, along with a refreshed terminal interface and a new VS Code extension, aims to enhance the coding experience. Additionally, the model includes context editing capabilities and a memory tool to manage long-running tasks without hitting context limits.

Developers can also create custom AI agents using the Claude Agent SDK, which provides the infrastructure used to build Claude Code. This SDK is available on the Claude Developer Platform, Amazon Bedrock, and Google Cloud's Vertex AI, with pricing remaining consistent with previous versions at $3/$15 per million tokens.

Beyond coding, Claude Sonnet 4.5 is positioned as a versatile chatbot for various workplace tasks, excelling in fields such as financial services, cybersecurity, and law. The model also includes improved defenses against prompt injection attacks, a significant cybersecurity risk, and is designed to be less prone to behaviors like sycophancy and power-seeking.