Anthropic Ships Claude Code Cloud Sessions so Developers Can Code Without a Laptop

Anthropic promotes Claude Code cloud sessions to general availability with one-time credits of $100 for Pro and $250 for Max users.

·
·
Anthropic Ships Claude Code Cloud Sessions so Developers Can Code Without a Laptop
Read4 min
TypeNews
  • Claude Code cloud sessions exit research preview and reach general availability on Anthropic-hosted infrastructure.
  • Pro subscribers get a one-time $100 credit, Max subscribers get $250, claimable through October 7.
  • Credit sits outside normal plan usage limits and auto-applies when a cloud session starts.
  • Launch a session from claude.ai/code, the mobile app, desktop app, or claude --cloud.
  • GitHub connection required since each session runs on its own branch and repo copy.
  • Enterprise teams can route sessions to self-hosted runners inside their own networks via partners like Coder.

Claude Code cloud sessions reach general availability

Anthropic has moved Claude Code cloud sessions out of research preview and into general availability. Developers can run coding tasks on Anthropic-hosted infrastructure, disconnect their computer, and return later to review or redirect the work.

Long-running refactors, migrations, dependency upgrades, and test loops no longer require the originating machine to remain awake and connected. Anthropic is also offering existing Pro and Max subscribers a one-time cloud credit, separate from their regular plan allowance.

Remote work that survives disconnects

A cloud session runs Claude Code on remote infrastructure and retains its state after the computer that launched it disconnects. Anthropic hosts sessions by default, while organizations with the self-hosted option can route them to infrastructure they control. Anthropic’s cloud-session documentation covers the supported workflow.

Developers can start sessions from four surfaces:

  • The web at claude.ai/code
  • The Code tab in the Claude mobile app
  • The Claude desktop app
  • The terminal with claude --cloud

After authenticating with claude auth login, a developer can send follow-up instructions from the Claude CLI on another machine. The remote session maintains its own state, which allows work to continue from a second laptop, mobile device, or authenticated automation host.

Credit comes with a deadline

Existing Pro and Max subscribers can claim one promotional credit per account through the Claude Code CLI with /claim-credit. Anthropic set October 7 as the claim deadline.

One-time cloud-session credits
Plan Credit Eligible usage
Pro $100 Cloud sessions only
Max $250 Cloud sessions only

Once claimed, the credit applies automatically when a cloud session starts. It remains separate from the account’s regular usage allowance, so an eligible session can continue against the promotional balance after the standard allowance has been exhausted. Cloud work continues under the account’s normal usage terms after the credit is spent.

Automation without an open laptop

Remote execution makes Claude Code more suitable for jobs that take longer than an interactive terminal session. A developer can assign a multi-file change, leave the session running, inspect its progress later, and provide additional instructions without returning to the original computer.

Claude Code routines can also launch cloud sessions from cron schedules, HTTP endpoints, or GitHub events. That combination supports recurring maintenance, issue-driven changes, and automated test-fix cycles without keeping a developer workstation online.

GitHub branches define the handoff

A connected GitHub account is required before starting a hosted session. Each session checks out its own copy of the repository and works on a separate branch, producing changes that developers can inspect before merging.

Concurrent sessions can edit the same files, so overlapping work may produce ordinary Git merge conflicts. Branch protections, required reviews, and CI checks remain relevant because the agent’s output enters the repository through the same review path as other code changes.

Self-hosting stays in beta

Organizations with source-residency or network-control requirements can run sessions on infrastructure inside their own environment through Anthropic’s public beta program. Repository checkouts, build artifacts, secrets, and modified files remain on machines provisioned by the organization.

Partners such as Coder provide self-hosted runners and policy controls for that deployment model. Anthropic labels hosted cloud sessions as generally available and the self-hosted path as a public beta, so teams should evaluate their support, retention, networking, secret-management, and audit requirements separately.

A crowded field and a temporary price break

OpenAI Codex, GitHub Copilot Workspace, Cognition Devin, and Google Jules offer variants of the same asynchronous workflow: assign a task, let an agent modify code in the background, and review the resulting branch or pull request.

Anthropic’s one-time credit lowers the cost of evaluating longer cloud runs and addresses a practical constraint in Claude Code: background work previously consumed the same allowance used for interactive sessions. The promotion provides temporary capacity; regular plan limits govern usage after the balance is depleted.

Test the workflow before wiring it into CI

A focused evaluation should use a noncritical repository and a task with clear acceptance criteria, such as a dependency upgrade, a multi-file refactor, or a test-fix loop. The basic sequence is:

  1. Connect the GitHub repository.
  2. Claim the promotional credit with /claim-credit.
  3. Launch a session with claude --cloud or another supported client.
  4. Review the generated branch, test results, and logs.
  5. Verify permissions, secret handling, network access, retention, and usage limits before production adoption.

General availability gives platform teams a clearer product-lifecycle signal than the former research preview. Production use still requires documented expectations for service availability, support, cost controls, repository access, and recovery when an agent run fails or produces conflicting changes.

Trending
  • No trending articles

Comments

avatar

Next Reads