Pricing
The English user guide is currently in beta preview. Most of the documents have been automatically translated from the Japanese version. Should you find any inaccuracies, please reach out to Flatt Security.
Takumi Runner uses usage-based pricing based on workflow execution time. This page explains the pricing model.
Billing Model
Takumi Runner pricing is based on execution time (per minute). Fractions of a minute are rounded up, and the minimum charge per job is 1 minute. Billing is monthly, and payment is made through the payment method registered with your Takumi subscription.
Price per SKU
The current SKU pricing is as follows:
| SKU | vCPU | Memory | Price (JPY/min) | Price (USD/min) |
|---|---|---|---|---|
linux-amd64-2core | 2 | 8 GiB | ¥1 | Contact us |
linux-amd64-2core corresponds to the runs-on: takumi-runner label. Billing in currencies other than JPY may be available. Please contact us for details.
What Is Billed
Billed Time
Billed execution time is measured based on timestamps from GitHub's workflow_job webhook. Specifically, the difference between started_at when the job starts (action: in_progress) and completed_at when the job completes (action: completed) is billed. The following diagram shows the billing interval within the job lifecycle:
queued in_progress completed
| | |
+------------------+------------------------+
| waiting (free) | billed (per minute) |
+------------------+------------------------+
^ ^
started_at completed_at
- Billed: Job step execution time (checkout, build, test, etc.)
- Not billed: Queue wait time, VM startup/shutdown time
Job execution time is charged to the billing month based on the completed_at timestamp.
Failed and Cancelled Jobs
If a job fails or is cancelled partway through, the actual execution time is still billed. Jobs cancelled before being assigned to a runner are not billed.