Announcing Claude 3.7 Sonnet on Snowflake Cortex AI

We are thrilled to announce that as part of our strategic partnership with Anthropic, Snowflake customers will now have access to Claude 3.7 Sonnet in Snowflake Cortex AI. 

Anthropic and Snowflake entered a multi-year partnership in November 2024 to help enterprises develop and scale easy, efficient and trusted AI products, starting with the launch of Claude 3.5. We are building on this momentum with Claude 3.7 Sonnet. The model will be available for customers to use within the secure Snowflake perimeter on Snowflake Cortex AI, a fully managed AI service that provides a suite of generative AI features.

Reasoning is an integrated capability of Claude 3.7 Sonnet. This model offers hybrid reasoning, allowing users to configure its response time based on their needs. For applications demanding low latency, it can answer quickly; for more complex agentic workflows, it can engage in more extensive thought. This hybrid approach enables enterprises to build a wide range of applications within a unified framework.

Snowflake Cortex AI

Snowflake Cortex AI is a suite of integrated features and services that include fully managed LLM inference, fine-tuning and RAG for structured and unstructured data, so that customers can quickly analyze unstructured data alongside their structured data and expedite the building of AI apps. The unified AI and data platform helps organizations turn AI concepts into reality in days. Enterprises of all sizes and industries can now accomplish a range of use cases from text summarization and sentiment analysis to the development of powerful AI assistants. 

Claude 3.7 Sonnet

Reasoning made practical

Anthropic Claude 3.7 Sonnet operates in both ‘standard’ mode and ‘extended thinking’ mode. Customers can choose the appropriate mode to get a quick response or deeper reasoning. In ‘extended thinking’ mode, the model self-reflects before answering and can provide a more comprehensive response, according to Anthropic. The model engages in self-reflection before answering, improving its performance. This makes it particularly beneficial for industries that rely on multi-step problem-solving and high-level analytics. The choice and control allow customers to make trade-offs between latency (cost) and response quality.

Improvements in coding capabilities

Claude 3.7 Sonnet shows significant improvements in coding and front-end web development, per Anthropic. Early testing and benchmarking data published by Anthropic show that Claude 3.7 outperforms its predecessor, Claude 3.5, across most coding benchmarks, making it a more capable model for development tasks. 

Beyond coding, Claude 3.7 Sonnet is also a leader in general reasoning and multimodal capabilities. It outperforms its predecessors in tasks that require reasoning across different domains and modalities. This flexibility makes it an ideal tool for businesses looking to integrate AI into diverse workflows, from customer service to data analysis.

Longer output context

Anthropic Claude 3.7 Sonnet offers up to 200K tokens in output. The longer output capacity allows users to retrieve more information from structured or unstructured queries. The reasoning capabilities, combined with longer context windows, make it well-suited for agentic workflows. 

By integrating Claude 3.7 Sonnet into Snowflake, users gain access to cutting-edge AI functionality that interacts with their data while maintaining enterprise-grade security, privacy and unified governance — all within Snowflake’s secure perimeter. Now, users can get to data-driven insights faster, creating exciting opportunities for businesses.  

Delivering responsible AI

According to Anthropic, the model is optimized for real-world tasks that better reflect how businesses use LLMs. Anthropic has conducted extensive testing to ensure it meets the standards for security, safety and reliability. Benchmarking results from Anthropic show that Claude 3.7 makes more nuanced distinctions between harmful and benign requests, reducing unnecessary refusals by 45% compared to its predecessor.

Looking ahead

Snowflake prioritizes security and privacy, and Anthropic is dedicated to building secure, reliable AI — creating strong alignment in delivering cutting-edge innovations. Claude 3.7 Sonnet, the first model with hybrid reasoning, is now available within the secure Snowflake perimeter. With this improvement in coding and focus on responsible AI, enterprises can now expedite the delivery of performant generative AI applications and data agents. 

Learn more 

• Get started with Cortex AI and Claude 3.7 — Quickstart guide

• Get a practical guide to AI agents — Download now

• Try Cortex Agents — Build your first Cortex Agent