# Anthropic: Claude 3.7 Sonnet Reasoning (High)

**Model:**

anthropic/claude-3.7-sonnet:thinking

**Max words:**

150,000

**Cost:**

6 per 100 words.

**Editor’s Choice**

**Capabilities:**

- Browsing

**Features:**

**Other features:**

## Anthropic: Claude 3.7 Sonnet Reasoning (High) Specialities

- Writing
- Coding
- Content
- Reasoning

## Anthropic: Claude 3.7 Sonnet Reasoning (High) Main Advantages

- Good programming skills
- Highly developed reasoning
- Good at developing complex problems
- Converting units of measurement

## Anthropic: Claude 3.7 Sonnet Reasoning (High) Model Limitations

- Code Hallucinations and Refactoring
- Overzealous Refusals and Ignored Instructions

Anthropic has designed **Claude Sonnet 3.7 Reasoning** with programmers in mind, creating its most advanced model to date. Initial testing confirms its leadership in software development, outperforming other models in handling complex codebases and building sophisticated applications.

In this release, we are including **Claude Sonnet 3.7** in two versions: **with and without Reasoning**. We’d love to hear your thoughts—Which one is your favorite? Should we keep both versions? Let us know!

### ⭐​ **Key Improvements of Claude Sonnet 3.7 Reasoning**

- **Enhanced Accuracy:** Claude Sonnet 3.7 Reasoning outperforms models like OpenAI O1, OpenAI O3 (High), and DeepSeek R1 in **SWE-bench Verified** tests, excelling in real-world software tasks.

- **Advanced Reasoning:** With self-reflection before responding, it significantly improves performance in **mathematics, physics, coding, and instruction following**.

- **Long-Form Output:** It tends to generate lengthy responses; for more concise results, be explicit in your instructions.