Claude on Google Cloud can now run on an Anthropic-managed stack

On October 23, Google Cloud made a second way to run Claude generally available: an Anthropic-managed deployment, alongside the existing Google-managed one.

Both options offer the same models, the same price, and the same Google Cloud bill, including eligible commitment drawdown. The split is who processes the data, and when new features arrive:

• Anthropic-managed: models and new platform features are available the day they launch, including Claude Managed Agents in beta, skills, and code execution. Anthropic is the data processor under its own commercial terms. Sign-in still uses Google Cloud IAM.
• Google-managed: the latest models are also available on day zero, but new platform features wait until they are built into Google's infrastructure. Google Cloud is the data processor under the existing customer agreement, with VPC Service Controls, customer-managed encryption keys, and data residency in the US, EU, and Asia-Pacific.

Google names Claude Fable 5.1, Sonnet 5.5, Haiku 5.5, and Opus 5.5 as available now.

For a company, this is two compliance paths on one cloud. Pick Anthropic-managed for the newest features. Pick Google-managed to stay on the existing contract and data-residency controls.