Anthropic Claude Opus 5 now available on Snowflake Cortex AI

Snowflake announced the public preview availability of Claude Opus 5, Anthropic's latest large language model, on Snowflake Cortex AI. As a launch partner with Anthropic, Snowflake is providing same-day access to the model within its secure computing perimeter, enabling customers to use Claude Opus 5 alongside existing Snowflake services including CoCo (the AI coding agent), Cortex Agents, Cortex AI Functions, Cortex Inference, and CoWork (a personal work agent for knowledge workers).

Claude Opus 5 is positioned as a thoughtful, proactive model that sets a new state of the art on coding and knowledge-work evaluations while delivering greatly improved performance over Opus 4.8. The model includes adjustable effort settings, allowing users to optimize for either intelligence or faster results by conserving tokens. It is designed to strengthen at verifying its own work and iterating until it succeeds, produces stronger visual outputs, and shows meaningful gains on scientific-research tasks, making it well suited to long-horizon agentic and analytical work.

Snowflake outlined five primary use cases for Claude Opus 5 on its platform. First, via CoCo (now generally available on macOS and Windows), developers gain an AI coding agent that understands enterprise data, governance, and workflows, with the model already loaded with their Snowflake catalog and RBAC roles active—eliminating cold starts and governance gaps. CoCo Desktop enables developers to turn natural language into production-ready pipelines, analytics, and AI agents. Second, CoWork serves as a personal work agent for knowledge workers, using Deep Research to decompose complex business questions across structured and unstructured data and return fully cited answers in minutes. Third, Cortex AI Functions allow analysts to apply natural language instructions directly in SQL using models such as Opus 5, enabling scalable AI pipelines across multimodal enterprise data. Fourth, Cortex Inference enables developers to build agentic enterprise applications within the Snowflake security perimeter, with support for the Anthropic SDK and integrations using Snowflake endpoints. Fifth, the combination of these capabilities enables enterprises to move agentic AI into production quickly, with agents handling complex, long-running tasks with greater consistency and reduced supervision.

The announcement emphasizes that all of this capability operates within Snowflake's security and governance perimeter, allowing enterprises to maintain data residency and compliance standards while adopting frontier AI reasoning. Customers can begin trying Claude Opus 5 in preview via the model picker in CoCo Desktop. The article includes forward-looking statements acknowledging that actual product offerings may differ from these announcements.

Snowflake's announcement reflects a strategic deepening of its partnership with Anthropic to embed frontier AI reasoning directly into enterprise data workflows. By offering Claude Opus 5 within Snowflake's security perimeter—rather than requiring data to leave the platform—Snowflake addresses a core constraint for regulated industries and large enterprises: the ability to run powerful AI agents without sacrificing governance, compliance, or data residency.

The launch positions Claude Opus 5 across multiple entry points in Snowflake's product suite, each targeting a different type of work. CoCo Desktop serves developers and data engineers automating infrastructure tasks; CoWork targets knowledge workers doing research and decision-making; Cortex AI Functions enable SQL-based analytics teams to embed AI reasoning into data pipelines; and Cortex Inference gives developers the flexibility to build custom agentic applications. This breadth means enterprises can adopt the same model across roles and use cases without switching tools or vendors.

The emphasis on adjustable effort settings—allowing teams to trade reasoning depth for speed and token efficiency—suggests Snowflake and Anthropic recognize that production agentic AI requires cost control alongside intelligence. Combined with Claude Opus 5's reported gains on long-horizon tasks and scientific work, the model appears designed for the sustained, multi-step reasoning that enterprise automation and research workflows demand.

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