Anthropic released self-hosted environments for Claude Code in public beta on August 6, 20261. Sessions started from web, mobile, desktop, or a routine run inside the organization’s own network, next to its internal services and toolchains, rather than on Anthropic-hosted infrastructure1. The approach uses long-lived “runners” that pick up sessions, in two modes: Fixed, which keeps a set number running, and On-demand, which starts and stops runners as sessions queue1. A runner can serve more than one session, but each session runs in its own checkout, so work stays isolated1.
Repository checkouts, build artifacts, secrets, and files a session creates or modifies stay on infrastructure the organization provisions. The conversation itself — prompts, responses, and tool results, which can include code that Claude reads — is sent to Anthropic for inference, and the session transcript is stored1. The beta is open to organizations on Claude Team and Enterprise plans; it is off by default and unavailable to organizations using ZDR1. Anthropic says it strongly recommends its hosted offering for most enterprises, and that teams choosing this route should plan to staff engineering for setup and ongoing maintenance1. The day before, on August 5, the company released an inline DLP feature for Claude Enterprise in beta.
Sources
- Run Claude Code sessions on your own compute - Anthropic official blog (August 6, 2026)