On October 22, 2026, Anthropic announced that “Claude Models on Google Cloud (Anthropic managed)” is generally available, “giving Google Cloud customers the full set of Claude Platform features.” In this option Anthropic operates the service and is the data processor, aimed at customers who want day-zero access to new beta and GA features. The existing Google-managed offering, with Google as data processor, remains for customers with strict regional data-residency needs or who want Google integrations such as Model Armor or Agent Engine.
Customers sign in with existing Google Cloud IAM credentials, and “Billing is through a single Google Cloud invoice and Claude usage draws down your existing Google Cloud commitment.” The feature list includes Managed Agents (beta), code execution, computer use, agent skills, the Files API, web search and fetch, the MCP connector (beta), prompt caching, batch processing and Claude Console access. Models offered are Claude Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5, initially on US and Global endpoints.
The same post states that Fast Mode on Claude Opus 5.5 is generally available, offering up to 2.5 times faster output token generation at $8 per million input tokens and $40 per million output tokens.
Why it matters: it lets Google Cloud customers spend committed cloud budgets on Anthropic’s own first-party platform rather than a reduced hosted subset. The day before, OpenAI announced its frontier models were coming to Google Cloud’s Model Garden, so in one week Google’s cloud became a sales channel for both of its main model rivals.
What it does not show: the post gives no regional availability beyond US and Global endpoints, and says existing discounts must be confirmed with an account executive and cannot be applied retroactively. The 2.5x fast-mode figure is Anthropic’s own “up to” claim.