Claude on Google Cloud now has an Anthropic-run option, with the same price

Google Cloud said on October 23 that an Anthropic-managed way to run Claude models is now generally available on its platform. In its announcement, Google says the main difference from the existing Google-managed option is who acts as the data processor.

Two ways to run Claude, one invoice

Michael Gerstenhaber, Google Cloud’s VP of product management for Gemini, wrote that the new option sits alongside the Google-managed one that already existed. Both give day-zero access to new Claude models, at the same pricing and on the same Google Cloud billing.

In the Anthropic-managed version, Anthropic runs the stack on Google Cloud infrastructure and is the data processor, under Anthropic’s Commercial Terms. Customers still sign in through Google Cloud IAM, keep Cloud Audit Logging and pay through their existing Google Cloud account. Spend can draw down eligible commitments. Google lists Claude Managed Agents (in beta), skills and code execution among the Claude Platform features available from day zero.

In the Google-managed version, Google Cloud is the data processor under existing customer agreements. Google lists VPC Service Controls, customer-managed encryption keys and data residency in the U.S., EU and APAC. It also names integrations with BigQuery and Model Armor. The trade-off is that new Claude Platform features arrive only after Google integrates them into its own infrastructure.

The models are the same either way. Google names Claude Fable 5.1, Sonnet 5.5, Haiku 5.5 and Opus 5.5.

The choice is about contracts, not capability

The comparison table in the announcement shows how narrow the technical gap is. Model availability is identical, and so is pricing. Only two rows differ: how quickly Anthropic’s own tools and betas appear, and whose terms govern the data.

That makes the decision mostly one for legal and compliance teams. A company that needs data residency in a specific region, customer-managed keys or a single contract with Google has a reason to stay on the Google-managed path. A team that wants Managed Agents and other new features the day Anthropic ships them has a reason to accept Anthropic as processor. Google does not say how much later features reach the managed-by-Google side, and gives no timetable.

The announcement also leaves several practical details open. It does not say which regions the Anthropic-managed option serves, or whether it offers the residency and encryption controls Google lists for its own version. It does not say whether the two options can be mixed inside one project. Those answers sit in the linked documentation pages, which the announcement points to but does not summarize. Statements about parity are Google’s own and have not been independently checked.

A busy week for Claude and Google Cloud

The timing fits a pattern. Anthropic recently shipped its first new API version since 2023, as we reported. According to that coverage, the new version, 2026-10-21, applies to the Amazon Bedrock and Google Cloud clients as well. Teams moving to the Anthropic-managed option should check their SDK versions against it.

Google also bets on breadth. A day earlier it announced that OpenAI’s GPT-6 models are coming to Google Cloud. That offer is a private preview for select enterprise customers, billed through Google Cloud Marketplace, with no per-token prices published. Per our reporting, OpenAI’s Azure exclusivity ended in April 2026 and its models reached Amazon Bedrock on April 28. Cloud providers are now competing on how many frontier models they carry and how those models are governed, not on exclusive access to one lab.

For Claude customers on Google Cloud the immediate effect is a new decision to make, not a new price. Anyone already running the Google-managed option sees no change unless they choose to move. Google points readers to a “Get started” link in the post for both paths.