Announcing powerful upgrades & a new pricing model for self-hosted runners

We’re excited to announce the release of Bitbucket Pipelines self-hosted runners v5.0. This release introduces powerful new capabilities, including the ability to customize resource allocations, take direct control of storage with Docker volume mounts, and store build data inside your own private cloud. We’re also setting the stage for future enhancements, such as APIs for managing your runners and the ability to run multiple, disposable macOS VMs on a single machine for scalable iOS/macOS development.

With this upgrade, we are also announcing the move to a paid licensing model. We recognize that self-hosted runners are a crucial requirement for teams to meet their compliance obligations. A paid model enables us to accelerate development, deliver enhanced support, and build the enterprise-grade experience you expect.

To allow time for transition, existing runners prior to v5.0 will remain available and free to use until Jun 3, 2026.

Upgrades included in v5.0

We’ve built v5.0 to give you more control over your CI/CD workflows by making it easier to manage your data, accelerate your builds, and tailor your pipelines to your team’s unique needs. Here’s what’s new:

Speed up builds with Docker volume mounts: Instead of downloading dependencies and rebuilding assets in every step, you can now share data between steps & builds by attaching persistent storage to your containers. For example, save build outputs or dependencies in one step and access them in the subsequent steps, reducing redundancy and speeding up your pipelines. Here’s an example:

pipelines:
  default:
    - step:
        runtime:
          self-hosted:
            volumes:
              - "/path/on/host:/path/in/container"
        runs-on:
          - linux
        script:
            - echo 1

Optimize resource allocation: Manage your resources by allocating CPU and memory based on what each step needs, all within your pipelines’ YAML. Choose from 1x–16x presets (e.g., 1x = 2 CPU cores, 16x = 32 cores), or fine-tune your allocations with custom step sizes for even more flexibility. Here’s an example of how you can set custom memory and CPU resources to a step.

pipelines:
  default:
    - step:
        runtime:
          self-hosted:
            memory: 6
            cpu: 2
        runs-on:
          - linux
        script:
          - echo 1

Control where your data is stored: Easily connect to S3 or Google Cloud Storage to keep caches & artifacts securely within your organization’s private cloud, giving you full control over where your data lives. Here’s an example:

pipelines:
  default:
    - step:
        oidc: true # default is true
        runtime:
          self-hosted:
            storage:
              aws:
                bucket: your-bucket-name
                region: your-aws-region # example, us-west-1
                oidc-role: my-role-arn
        runs-on:
          - linux
        artifacts:
          - file.txt
        script:
            - echo 1 > file.txt

Coming soon:

• Ephemeral macOS runners: Use Tart, an industry-leading macOS virtualisation framework to run disposable macOS VMs for iOS/macOS builds. This eliminates the need for dedicated hardware per agent and supports scalable CI for iOS/macOS development.

Your feedback drives our roadmap, so join the discussion on our Atlassian Community and help us shape what comes next.

Pricing

Our new pricing model is designed to be flexible, predictable, and easy to manage.

• Licensing per concurrent build slot: Self-hosted runners v5.0 and above will be licensed based on the maximum number of concurrent builds that can be executed across your workspace’s pool of runners.
• $15/month per concurrent build slot: Pricing per build slot provides significant value when compared to alternative commercial self-hosted CI tools. For most customers migrating from Bamboo, the costs of running a fleet of self-hosted runners will be significantly lower when ran on Pipelines.
• Plan inclusions: Standard tier workspaces include one free self-hosted runner slot, and Premium tier workspaces include two free slots.
• Configuring maximum concurrent builds: Customers on Standard or Premium plans can manage their allocation by setting the maximum number of licensed build slots in their workspace settings. If you have more builds than the maximum number of slots, they will be queued until slots become available. This feature is only available on Standard and Premium plans.
• Billing: You will be billed based on the maximum number of concurrent runners that you configure in any given month. Runners will be billed as an add-on to your Bitbucket Cloud subscription, starting on your next billing cycle.

Transition timeline and details

We understand that this is a significant change, and we are committed to ensuring a smooth transition.

• Transition Period: You can continue using runner versions prior to v5.0 for no charge until June 3, 2026 (monthly plans) or December 3, 2026 (annual plans). After these dates, we’ll disable scheduling steps on these runners.
• Migration: Customers can run v5.0 runners along with prior runner versions during the transition period. Pipelines will automatically manage distributing jobs to all available runners, with no manual input or intervention required from customers.
• Maximum concurrent builds applies only to 5.0 runners: Only builds executing on runners v5.0 and above will consume concurrent build slots. Builds executed on v3 runners for example, will not consume concurrent build slots.
• Runner version management: The updated runner management UI will clearly indicate runner versions and allow you to select specific runner versions when creating a runner.

Next steps

Before June 3, 2026:

• Configure your workspace concurrency limit to meet your organization’s requirements.
• Migrate your self-hosted runners to v5 in order to use the latest available features and upgrades.

Have questions or feedback?

Join the conversation on our Atlassian Community page, review our support docs or reach out to our support team.