Claude 3.7 Sonnet launches with the power of ‘extended thinking’

The development of AI language models has entered a new phase with increasingly sophisticated reasoning capabilities and specialized tools for developers. Anthropic’s latest release represents a significant advancement in how AI models process complex problems and assist with software development tasks.

Key Innovation: Anthropic has introduced Claude 3.7 Sonnet, featuring “extended thinking” capabilities that allow the AI to demonstrate its reasoning process step by step while solving problems.

• The model represents the first “hybrid reasoning model” in the market, letting users choose between quick responses or detailed chain-of-thought processing
• Developers can specify the number of tokens used for thinking, up to a 128,000 token output limit
• The extended thinking feature is available on all paid subscription plans

Technical Specifications and Pricing: Claude 3.7 maintains Anthropic’s existing pricing structure while introducing new capabilities.

• API costs remain at $3 per million input tokens and $15 per million output tokens
• Thinking tokens are included in the output pricing
• The model shows 45% fewer unnecessary refusals compared to previous versions, making it more responsive to user requests

Performance and Benchmarks: The new model demonstrates particular strength in programming-related tasks.

• Claude 3.7 achieved top scores on SWE-bench Verified, which tests AI handling of real-world software issues
• The model excelled in TAU-bench, which evaluates AI agents on complex tasks with user and tool interactions
• Enhanced GitHub integration is now available across all Claude plans

Claude Code Introduction: Anthropic has launched its first AI agent specifically designed for developers.

• The tool operates through a console terminal as an autonomous coding assistant
• It can search codebases, manipulate files, write tests, and manage GitHub repositories
• Internal testing shows Claude Code completing tasks in single sessions that typically require 45+ minutes of manual work
• The tool is currently available as a limited research preview

Availability and Access: The new capabilities are being rolled out across multiple platforms.

• Claude 3.7 Sonnet is accessible through the Claude website, mobile app, Anthropic API
• The model is also available through Amazon Bedrock and Google Cloud’s Vertex AI
• The basic subscription remains at $20/month for Claude Pro

Looking Ahead: While Claude 3.7 Sonnet represents a significant advance in AI reasoning capabilities, the limited nature of current subscription plans may pose challenges for power users, particularly developers who require extensive usage. The success of Claude Code’s research preview and user feedback will likely influence future developments in AI-assisted software development tools.