You can now try Anthropic's Claude 3.7 Sonnet on Google Vertex AI & Amazon Bedrock

Google Vertex AI, the tech giant’s AI platform, is welcoming a new addition to its models catalog: Anthropic’s Claude 3.7 Sonnet.

Google Cloud recently announced Anthropic’s hybrid reasoning model on Vertex AI. This model combines rapid responses and extended reasoning, with improvements for real-world applications, including coding.

The model comes with two variants, the original and the “extended thinking mode,” which is available “on all surfaces except the free Claude tier.” It is also available to try on Amazon Bedrock, among other AI platforms.

Claude 3.7 Sonnet’s pricing stays the same as previous versions, costing $3 per million input tokens and $15 per million output tokens, including thinking tokens.

But what’s so different about these modes? In regular mode, the model gives quick answers. In extended mode, it takes more time to think things through and give more detailed, careful answers, which is especially helpful for harder tasks like math or coding and is similar to Copilot’s Think Deeper.

Other than that, tools like Claude Code are also available for developers alongside enterprise-grade security features.

“Claude Code lets developers delegate coding tasks to Claude directly from their terminal and is available through Anthropic’s limited research preview,” Google describes.

Anthropic says that Claude 3.7 Sonnet has demonstrated strong performance in coding, excelling in benchmarks like SWE-bench Verified and TAU-bench, which test AI models on real-world software tasks and complex problem-solving. It achieved top scores, particularly in coding tasks involving user interactions and tool usage.