Anthropic’s New Claude 3.7 Sonnet Will “Think” for as Long as You Want

Key Takeaways

• On Monday, Anthropic released the latest Claude 3.7 Sonnet that can think for as long as you want.
• Users will be able to choose between instant responses or a step-by-step thinking process.
• The new AI is available to all users, both free and paid. However, the extended thinking feature is only for paid users.

OpenAI's biggest rival Anthropic has launched its smartest model so far - Claude 3.7 Sonnet. It's a hybrid model that can respond to your queries instantly or go through a detailed step-by-step thinking process, depending on what the user wants.

In a blog post, the company announced the launch of Claude 3.7 Sonnet and added that it will be available to all users, both free and paid. However, the extended thinking feature is only limited to paid users for now.

What Makes Anthropic's Reasoning Model So Special?

Anthropic isn't the first AI company to create a model that can think. OpenAI, DeepSeek, and xAI have done it before. But the difference is other companies position reasoning models as the future of AI whereas Anthropic views it as an addition alongside the existing model.

This way, if you have a simple question like Who was the first female president?", you'll get a quick and direct answer whereas if you have a complicated math problem, you'll get a detailed answer with all the steps to help you understand the process better.

What we're really trying to do is make it really seamless to adopt this capability where it makes sense, but not have it brought to bear where it doesn't make sense - Mike Krieger, Anthropic's chief product officer

The second benefit is if you are using Claude 3.7 through API, you can customize the number of tokens it uses for each thinking process. For example, you can instruct the tool to think in no more than 100 tokens. The fewer tokens it uses, the faster the response will be.

Note: Using fewer tokens is more cost-effective because paid plans for such AI tools are often priced per number of tokens.

Speaking of the pricing, Anthropic has announced that the cost will remain the same for both standard thinking and extended thinking i.e. $3 per million input tokens and $15 per million output tokens which includes the thinking tokens.

Who Is Claude 3.7 Sonnet Best For?

Anyone can use this tool but we feel that it's best for organizations that require extensive coding on a day-to-day basis. Early testing has shown that Claude is one of the best tools in the market to handle complex coding tasks, especially in planning code changes and dealing with full-stack updates.

It's also one of the few tools to create production-ready codes with impeccable design taste and negligible errors. The best part is that the testing has been done by third parties like Cursor, Cognition, and Canva. So you can be sure that the results are completely unbiased.

Read More: Anthropic CEO says human-level AI models might arrive by 2026

Claude Code

In addition to Claude 3.7 Sonnet, Anthropic has also announced the launch of Claude Code - the first Anthropic agentic tool, in a limited research preview.

This tool can search and read codes, edit files, write and run tests, and push codes to GitHub and the user will be kept in the loop for all of these stages.

In early testing, it was found that tasks that manually take 45+ minutes to be done can be completed within minutes with the help of Claude Code.

The tool is in the early stages hence developers are encouraged to join the preview and share their feedback to help create the ultimate AI coding tool for the future.

The post Anthropic's New Claude 3.7 Sonnet Will Think" for as Long as You Want appeared first on Techreport.