Claude 3.7 Sonnet is designed to handle complex tasks, such as coding and agentic tasks, with improved accuracy and fewer refusals compared to previous models. It features a "visible scratch pad" to display its internal planning process, although some parts may be redacted for safety. Alongside Claude 3.7 Sonnet, Anthropic is launching Claude Code, a tool for developers to execute coding tasks directly from the terminal. This release comes as AI labs rapidly develop new models, with Anthropic aiming to lead the industry while maintaining a focus on safety.
Key takeaways:
- Anthropic has released Claude 3.7 Sonnet, a hybrid AI reasoning model that can provide both real-time and more considered answers, with reasoning features available to premium users.
- Claude 3.7 Sonnet is more expensive than other reasoning models like OpenAI's o3-mini and DeepSeek's R1, but it offers hybrid capabilities that combine real-time and reasoning functions.
- The model has shown improved performance in real-world tasks, scoring higher than competitors in tests like SWE-Bench and TAU-Bench, and it reduces unnecessary refusals by 45% compared to its predecessor.
- Alongside Claude 3.7 Sonnet, Anthropic is launching Claude Code, an agentic coding tool that allows developers to run tasks directly from their terminal, initially available to a limited number of users.