"Claude 3.7 Sonnet" - "The world's first hybrid reasoning AI model supporting both standard and extended thinking modes."

## Reasoning Modes of Claude 3.7 Sonnet Claude 3.7 Sonnet supports two reasoning modes: **standard thinking** for fast responses and **extended thinking** for advanced, step-by-step reasoning. ## Areas of Excellence for Claude 3.7 Sonnet Claude 3.7 Sonnet excels in **code and front-end development**, **mathematics**, and **science**. It also performs exceptionally well in **instruction following**, **general reasoning**, **multimodality**, and **agent coding**. ## Performance in Pokémon Gameplay Tests Claude 3.7 Sonnet outperforms all previous models in **Pokémon gameplay tests**, demonstrating its advanced reasoning capabilities. ## Platforms Supporting Claude 3.7 Sonnet Claude 3.7 Sonnet is available on **Claude plans** (Free, Pro, Team, Enterprise), **Anthropic API**, **Amazon Bedrock**, and **Google Cloud’s Vertex AI**. ## Pricing Structure of Claude 3.7 Sonnet The pricing for Claude 3.7 Sonnet is **$3 per million input tokens** and **$15 per million output tokens**. The extended thinking mode is available on all plans except the Free tier. ## Additional Tool Associated with Claude 3.7 Sonnet **Claude Code** is an additional tool associated with Claude 3.7 Sonnet. It is a command line tool for agentic coding, available in a limited research preview. ## Key Features of Claude 3.7 Sonnet Key features of Claude 3.7 Sonnet include: - Support for **standard thinking** and **extended thinking** modes. - Exceptional performance in **code and front-end development**. - Significant improvements in **mathematics and science** with extended thinking. - Superior performance in **Pokémon gameplay tests**. - Broad availability on multiple platforms. - Transparent pricing structure. ## Purpose of Claude Code **Claude Code** is designed to enhance the utility of Claude 3.7 Sonnet for developers by enabling tasks such as **code search**, **file editing**, **writing/running tests**, and **GitHub integration**. ## Safety Measures for Claude 3.7 Sonnet Safety measures for Claude 3.7 Sonnet include extensive testing with external experts to meet **security**, **safety**, and **reliability** standards. The model addresses **prompt injection attacks** and evaluates the **trustworthiness** of its reasoning. ## Benchmarks for Claude 3.7 Sonnet Claude 3.7 Sonnet has been evaluated against **SWE-bench Verified** and **TAU-bench**, achieving state-of-the-art results. It also outperformed previous models in **Pokémon gameplay tests**. ### Citation sources: - ["Claude 3.7 Sonnet"]("https://www.anthropic.com/news/claude-3-7-sonnet") - Official URL Updated: 2025-03-26