We’re excited to announce that Anthropic Claude 3.7 Sonnet is now natively available in Databricks across AWS, Azure, and GCP. For the first time, you can securely access Claude’s advanced reasoning, planning, and agentic capabilities directly within Databricks. By combining Claude with the Databricks Data Intelligence Platform, enterprises can build domain-specific AI agents on their unique data—with full governance and observability across the entire data and AI lifecycle.
Claude 3.7 Sonnet is Anthropic’s most advanced AI model to date, delivering best-in-class performance on complex tasks like agentic reasoning and multi-step planning, multi-turn conversations with long context and deep document and data understanding. It’s the first hybrid reasoning model—allowing developers to control its “thinking time” and inspect its reasoning steps, a key capability for building transparent, trustworthy AI agents. Benchmarks have demonstrated Claude's leadership in complex multi-step agentic systems and coding capabilities, making it ideal for operational agents in enterprise environments.
As the first hybrid reasoning model on the market, users can control how long the model can think for, and visibility inspect the model’s reasoning.
Databricks users can now access Claude 3.7 natively and securely across AWS, Azure, and GCP—no setup required. Mosaic AI Gateway automatically governs usage with built-in logging, guardrails, and PII detection, so teams can confidently use Claude 3.7 Sonnet like any other model on Databricks.
This launch marks the beginning of a strategic partnership between Anthropic and Databricks. Together, we’re enabling organizations to build domain-specific AI agents that can reason deeply over their proprietary data. Here are two key ways to get started:
Claude’s advanced reasoning, paired with Mosaic AI, provides everything teams need to develop, deploy, and scale enterprise-grade agents.
Claude's capabilities also make it ideal for advanced batch inference tasks that require high-quality reasoning.