Claude Models on Google Cloud (Anthropic Managed) Reaches General Availability
Claude Models on Google Cloud (Anthropic managed) reached general availability on October 22. Customers access Anthropic's Claude API through a Google Cloud project with Google Cloud authentication and billing, while Anthropic operates the API and model infrastructure. US-only inference is a separate request setting, and the endpoint label alone does not make processing Google Cloud-only.
Claude Models on Google Cloud (Anthropic managed) is generally available, Anthropic announced on October 22. Customers access the Claude API through a Google Cloud project using Google Cloud credentials and billing, but Anthropic operates the API and its underlying platform on Anthropic systems. (Anthropic's GA announcement; Anthropic's platform documentation)
The service is aimed at teams that want Anthropic's Claude API and platform features through a Google Cloud account. It is not the same deployment boundary as Google-managed Claude, and the two options differ in who operates the service, the API surface, feature availability, and processor responsibility. (Google Cloud's comparison)
What the GA offering includes
Anthropic's launch announcement lists code execution, computer use, Agent Skills, the Files API, web search and web fetch among the available features. The announcement marks Claude Managed Agents, Advisor strategy, and an MCP connector as beta features. It says new features and betas ship on the Anthropic-managed offering the same day they go live on the native Claude Platform. That timing does not guarantee access to every organization-limited beta: Anthropic's technical docs say a beta limited to selected Claude API organizations requires the new Anthropic organization to have the same access. The technical documentation also cautions that its feature list is not exhaustive and identifies unsupported features, including API keys, MCP tunnels, memory stores on self-hosted environments, and pre-built self-hosted sandbox workers. (Anthropic's GA announcement; Anthropic's feature documentation)
Google Cloud IAM is used for access, and billing runs through Google Cloud Marketplace rather than a separate Anthropic invoice. Anthropic says Claude usage draws down an existing Google Cloud commitment; Google's documentation makes that drawdown subject to its Marketplace Commitment Drawdown Policy. Teams with an existing discount are advised to confirm it before starting because Anthropic says discounts cannot be applied retroactively. (Anthropic's GA announcement; Anthropic's billing documentation; Google Cloud's documentation)
Understand the processing boundary
Google Cloud and Anthropic act as separate, independent data processors for their respective parts of the service. Google handles the integration layer, including IAM authentication, API routing, Cloud Audit Logs, and Marketplace billing; Anthropic operates the Claude API and underlying platform as a processor for its part. (Google Cloud's processor and responsibility documentation; Anthropic's service documentation) Anthropic's documentation also says some supporting services, including its trust and safety systems, process data outside Google Cloud. (Anthropic's service documentation)
Google states that data storage and inference occur on Anthropic-managed infrastructure, which may route outside Google Cloud. Its documentation also cautions that subservices may change without notice. If an organization requires Google Cloud to be the sole data processor or requires data and inference to remain inside Google's security boundary, Google recommends its Google-managed Claude option instead. (Google Cloud's documentation)
Endpoint geography and inference are separate
Anthropic's announcement lists US and Global endpoints, but its technical docs say to leave the API location at global and set inference_geo: "us" on each request when US-only inference is required. The data-residency documentation says inference_geo is supported on Claude 4.6 and later; US-only inference on those models is priced at 1.1 times the standard rate across token pricing categories. For Claude Haiku 5.5 prompts over 100,000 tokens, the multiplier also applies to the higher long-context prices. (Anthropic's GA announcement; Anthropic's managed-service documentation; Anthropic's data-residency documentation)
Inference geography is not the same as workspace geography. Anthropic says workspace geography controls where data is stored at rest and where endpoint processing occurs; for this Google Cloud offering, workspace geography is fixed to US and cannot be configured. (Anthropic's data-residency documentation)
Plan the move as a separate organization
Activating the service creates a new Anthropic organization for the Google Cloud project. Google says API keys, workspaces, HIPAA controls, Zero Data Retention arrangements, and settings from an existing first-party Anthropic organization do not carry over; Anthropic says ZDR is available on request for the new organization. (Google Cloud's setup documentation; Anthropic's organization and retention documentation)
| Decision area | Anthropic-managed | Google-managed |
|---|---|---|
| Service operator | Anthropic operates the Claude API and platform; storage and inference use Anthropic-managed infrastructure. (Google Cloud's comparison) | Google operates the stack and is the sole processor for the Google-managed service. (Google Cloud's comparison) |
| Billing | Google Cloud Marketplace billing in Claude Consumption Units; usage can draw down on a Google Cloud commitment subject to Marketplace policy. (Google Cloud's billing documentation) | Google Cloud Agent Platform billing. (Google Cloud's billing documentation) |
| Best fit | Teams seeking Anthropic-operated Claude API features through Google Cloud access and billing. (Anthropic's GA announcement) | Teams that require Google as sole processor, Google Cloud security-boundary controls, regional residency, or built-in Google Cloud integrations. (Google Cloud's comparison) |
The key decision is not just which cloud account sends the request. Confirm which company processes each part of the service, where data is stored and inference runs, which organization-level controls must be recreated, and whether US-only inference affects cost.
Read next
Related coverage- OpenAI
OpenAI Models and Codex Are Coming to Google Cloud
OpenAI and Google Cloud plan to make OpenAI frontier models and Codex agentic workflows available in Google Cloud's Model Garden. The announcement says frontier models can be purchased through Google Cloud Marketplace and general availability is planned in the coming weeks, but it does not name the models or a launch date.
- Claude
Claude Makes Enterprise-Managed Auth for MCP Connectors Generally Available
Anthropic says enterprise-managed auth for MCP connectors is now generally available. Claude Team and Enterprise admins can centralize authorization through their identity provider, while users receive connector access automatically without individual OAuth consent flows.
- Claude
Claude Opens a Plugin Directory Submission Portal for MCP and Skills
Anthropic launched a Claude directory submission portal on September 25, 2026. Developers on paid Claude plans can submit a remote MCP connector or a GitHub-hosted plugin bundle that packages MCP servers and skills, follow validation and safety review, publish approved versions, and see listing and usage analytics.
- Claude
Claude Managed Agents Adds a Terminal Session Viewer and Auto Permission Mode
Claude Managed Agents now includes a terminal workflow for attaching to live sessions and a local web session viewer. Anthropic also added an auto permission mode that evaluates each agent or MCP tool call against the intent in user.message and runs, denies, or pauses for approval.
Sources
Frequently Asked Questions
What reached general availability on Google Cloud?
Claude Models on Google Cloud (Anthropic managed), which lets customers access Anthropic's Claude API through a Google Cloud project using Google Cloud authentication and billing, was announced as generally available on October 22, 2026.
Does a US endpoint mean data stays inside Google Cloud?
No. Anthropic operates the API on its systems, and Google says storage and inference occur on Anthropic-managed infrastructure that may route outside Google Cloud. The inference_geo setting can pin inference to US infrastructure, but Google-managed Claude is the option for Google Cloud as sole processor and for data and inference inside Google's security boundary.
Do existing Anthropic organization settings carry over?
No. Setup creates a separate Anthropic organization. API keys, workspaces, HIPAA controls, Zero Data Retention arrangements, and settings from an existing first-party Anthropic organization do not carry over.