Anthropic, backed by Amazon, has launched Claude 4, its next-generation AI model, on May 22, 2025, setting new benchmarks in coding, reasoning, and AI agent capabilities. The new model is available to paying Claude subscribers and offers enhancements for both free and paid users.

Anthropic's latest AI model, Claude 4, introduces two versions: Opus 4 and Sonnet 4. Claude Opus 4 excels as the world's leading coding model, demonstrating consistent performance in complex and long-term tasks, while Claude Sonnet 4 significantly enhances coding and reasoning abilities, responding more accurately to instructions. Both models now support tool use, enabling web searches during extended thinking to improve response quality. They can also use tools in parallel, follow instructions with greater precision, and exhibit improved memory through local file access, extracting crucial information to maintain continuity.

Mike Krieger, Chief Product Officer at Anthropic, noted that Claude 4 Opus can operate agentically in tasks like playing Pokémon for up to 24 hours, a significant improvement from the previous 45-minute limit of Claude 3.7 Sonnet.

This release marks a shift in strategy for Anthropic, which has focused on enhancing Claude's proficiency in complex tasks rather than investing in chatbots. The new API capabilities, including code execution tools, MCP connector, files API, and prompt caching, are designed to empower developers in building more robust AI agents.

Experts predict that these advancements will further integrate AI into professional workflows, facilitating seamless pair programming and more efficient task management across various industries.