Anthropic Claude Opus 5 launches on Snowflake Cortex AI

Snowflake announced same-day availability of Claude Opus 5, Anthropic's latest AI model, in public preview on Snowflake Cortex AI, a secure platform for building AI applications within an enterprise's data environment. As a launch partner with Anthropic, Snowflake is providing immediate access to Opus 5 across its AI tooling.

Claude Opus 5 represents an advancement over its predecessor, Opus 4.8, setting a new state of the art on coding and knowledge work evaluations. The model is described as thoughtful and proactive, with performance approaching Claude Fable 5 on frontier intelligence tasks. Key improvements include stronger verification and iteration capabilities—the model can check its own work and refine answers until they succeed—stronger visual reasoning outputs, and meaningful gains on scientific-research tasks. The model features adjustable effort settings, allowing developers and users to optimize either for maximum intelligence or for faster results and lower token consumption.

Snowflake is integrating Opus 5 across five entry points. CoCo, Snowflake's AI coding agent, now comes in a desktop version generally available on macOS and Windows. CoCo Desktop provides developers an integrated workspace with access to local files, repositories, terminals, and Snowflake, with the developer's RBAC (role-based access control) roles already loaded—eliminating cold-start delays and re-explaining data models. With Opus 5, CoCo Desktop enables improved reasoning and code generation for complex, long-running agentic tasks; the article claims it can turn natural language into production-ready pipelines and analytics, collapsing weeks of glue work into minutes while maintaining governance. CoWork, a personal work agent for knowledge workers, uses Opus 5 to enhance reasoning and consistency across multi-step workflows. With Deep Research, CoWork can decompose complex business questions across enterprise data and return fully cited answers in minutes; with proactive automations, it can surface insights and recommend actions unprompted. Cortex AI Functions allow analysts to apply natural language instructions directly in SQL using Opus 5, for example by invoking an AI_COMPLETE function to review financial filings and summarize revenue trends and risk factors. Cortex Inference enables developers to build agentic enterprise applications within Snowflake's security perimeter, with Opus 5's adjustable effort levels balancing reasoning depth and latency per task; developers can integrate using the Anthropic SDK by pointing it to their Snowflake Cortex Inference endpoint.

The announcement emphasizes the value of keeping data and AI processing within Snowflake's security and governance perimeter. Enterprises can build and deploy agentic AI—systems capable of handling complex, long-running tasks with reduced supervision—without moving sensitive data outside their organization. Snowflake frames this as enabling customers to accelerate their path to production-ready agentic AI at scale.

Snowflake's partnership with Anthropic marks a strategic move to embed advanced reasoning capabilities directly into enterprise data workflows. By positioning Claude Opus 5 within Cortex AI—a secure compute environment that sits atop Snowflake's data platform—the company enables customers to build agentic systems (AI agents that autonomously plan and execute multi-step work) without ever exporting sensitive data. This addresses a core enterprise concern: the tension between using cutting-edge AI and maintaining data governance.

The announcement emphasizes several entry points for adopting Opus 5. CoCo Desktop, now generally available, transforms how data engineers and analysts work by collapsing manual pipeline and glue-work into minutes of natural-language instruction—a concrete productivity claim the body supports. CoWork, Snowflake's personal work agent, extends the same capability to broader knowledge workers, allowing them to decompose complex business questions across structured and unstructured data and receive cited answers. Cortex AI Functions make it easy to invoke Opus 5 directly from SQL queries, further lowering the friction for enterprises to embed AI reasoning into existing analytics workflows.

The model's adjustable effort settings—allowing developers to trade reasoning depth for faster, cheaper results—address a practical deployment concern: not every task requires maximum inference compute. This flexibility, combined with the ability to access Opus 5 via the familiar Anthropic SDK pointed at a Snowflake endpoint, is designed to accelerate adoption among developers already familiar with Anthropic's tooling.

No comments yet. Be the first to share your thoughts!