PANGEA / GETTING STARTED

From repository to result.

Pangea is being designed for self-serve CI on AMD Ryzen 9 9950X hardware. Submit work, follow your jobs, and track your included minutes.

1. Connect your project

Open the console, continue with the demo GitHub sign-in and select a sample repository. The planned live flow will request access only to repositories you select. GitHub Actions compatibility is still in development; no public runner label or integration command is available yet.

2. Choose the resources for each job

Standard provides 4 vCPU and 8 GiB; Performance provides 8 vCPU and 16 GiB. These are proposed per-job allocations on shared Linux x86-64 hosts. Each vCPU is a logical CPU thread, not a whole physical core. You do not rent a complete node.

3. Start free and run a build

The planned trial gives you $3 credit for 14 days with no card. The demo’s “Run sample job” plays a scripted build, then shows logs, execution time and cost. Its 3-minute duration is illustrative, not a measured performance result. Demo progress resets on reload.

Submit → Queued → Running → Succeeded / Failed
                         ↳ Canceled / Timed out

Queue time: no charge
Execution: metered per second
Results: logs + exit status + artifacts

4. Check usage and enable billing

Open “Usage & billing”, choose a fixed monthly plan or optional capped usage billing, then continue to checkout. Confirming the demo activates a simulated plan without sending anything to Stripe. The live design is an online checkout through OND, with no sales call needed.

5. Keep control of your plan

Fixed plans show remaining included minutes and keep the compute price fixed. Standard uses one included minute per minute of execution; Performance uses two. Jobs pause at the allowance limit, with no automatic overage billing. The billing tab includes a demo cancellation action: it stops the next renewal while preserving access until period end.

6. Compare your own workflow

Before choosing a runner in the live service, compare execution time and end-to-end turnaround on the same commit, dependencies and cache conditions. We have verified the Ryzen 9 9950X hardware; comparative CI benchmarks are not published yet. CPU specifications alone do not promise a particular speedup.

Deploy applications with your agent

Pangea Apps is the second product preview: a fixed monthly workspace for multiple public apps. Your agent would submit a project, check deployment status and receive its public HTTPS URL. The proposed CLI command is pangea deploy . --name my-app; no CLI is distributed yet.

Running apps share the workspace’s CPU and memory allocation. The preview shows resource admission, sample logs, public URL previews and stop/start controls. App hosting plans are separate from CI execution plans. Explore Pangea Apps →

Before live access opens

Customer workloads will require validated job isolation, credential scoping, network separation, resource controls and cleanup. Self-serve registration, dispatch, durable usage metering and billing are still being built. This preview does not execute external code on our infrastructure.