This new generation coding model excels in real-world software development, using benchmarks like SWE-bench Verified to demonstrate state-of-the-art performance. Claude Sonnet 4.5 has showcased its ability to maintain focus on multi-step coding tasks for over 30 hours, greatly enhancing productivity and coherence across extensive codebases. It offers parallel tool execution, enabling simultaneous actions, and significantly improves performance on complex reasoning and domain-specific knowledge areas including finance, law, medicine, and STEM fields. Early customer feedback highlights notable advances in debugging, architecture design, and code comprehension, enhancing software development velocity and risk reduction.
Alongside its coding expertise, Claude Sonnet 4.5 is the most aligned model Anthropic has released, reducing harmful behaviors such as deception and delusion through extensive safety training and technical defenses against prompt injection attacks. Its agentic capabilities are further enhanced with the Claude Agent SDK, allowing developers to build their own capable AI agents using the same infrastructure powering Claude Code. For developers and users, Claude Sonnet 4.5 is available via the Claude API with pricing set at $3 per million tokens for the base plan and $15 for the premium plan. Additional features include code execution, file creation, and extended context handling, fostering enhanced performance in diverse AI-driven applications.