Anthropic Launches Claude Opus 4, Sonnet 4 AI Models for Advanced Coding, Long-Duration Tasks

Anthropic has officially launched its latest large language models, Claude Opus 4 and Claude Sonnet 4, marking a significant advancement in AI capabilities, especially for coding and long-context reasoning. Opus 4 is the company’s most powerful model, capable of sustaining complex, long-running tasks over thousands of steps, and is designed to power autonomous AI agents operating for extended periods. Sonnet 4 offers a speed and cost balance, while both models support extended thinking with tools such as web search and code execution during reasoning. These models are immediately available on platforms including Anthropic's API, Amazon Bedrock, and Google’s Vertex AI, with Sonnet 4 accessible for free and Opus 4 available for paying users. Internal testing revealed substantial improvements in memory training, allowing the models to manage longer tasks without heavy reliance on context windows, enabling them to write, summarize, and avoid getting stuck during complex workflows. Anthropic has been cautiously rolling out these models, having conducted early access tests under strict safety and usage limits, with the public launch aligning with their mid-year release tradition.