Anthropic unveils Claude 3.7 Sonnet, the smartest and first hybrid reasoning model — "Self-reflecting like humans," but with a trade-off on speed for quality AI responses

Anthropic recently unveiled Claude 3.7 Sonnet AI, touting it as the "most intelligent model to date" and the "first hybrid reasoning model on the market." According to the company, the new model generates near-instant or extended responses to queries.

Users can see the model's step-by-step thinking process, while its API can control how long Claude 3.7 Sonnet thinks before responding to a query. The model also ships with significant enhancements across coding and web development.

Aside from the new hybrid reasoning model, Anthropic is also launching Claude Code. As the name suggests, it's a command line tool for agentic coding. However, it's only available in preview and allows developers to assign tasks to the AI model directly from a terminal.

What's all the hype about Claude 3.7 Sonnet?

Claude 3.7 Sonnet with extended thinking - YouTube

Watch On

Last year, a report emerged indicating that top AI labs, including OpenAI, Anthropic, and Google, may be unable to develop advanced AI models due to a lack of high-quality content for training.

OpenAI CEO Sam Altman disputed the claims, indicating "there's no wall," echoing former Google CEO Eric Schmidt's about the lack of evidence indicating scaling laws have begun. AI labs are seemingly inclined and leaning more on reasoning models than previous iterations that heavily relied on online resources amid rising copyright infringement concerns.

This is the case with Anthropic's latest model. Claude 3.7 Sonnet ships with a drop-down menu that allows you to select the thinking mode you'd like to use depending on your query's complexity.

For instance, you can select Normal mode, which is the basic option often "best for most use cases." However, there's an Extended Mode option for complex queries, which best handles coding and math challenges.

“We've developed Claude 3.7 Sonnet with a different philosophy from other reasoning models on the market,” Anthropic added. “Just as humans use a single brain for both quick responses and deep reflection, we believe reasoning should be an integrated capability of frontier models rather than a separate model entirely. This unified approach also creates a more seamless experience for users."

It's worth noting that the new model might take a tad longer (up to a few minutes) to generate a response, especially when dealing with complex tasks or queries.