# GitHub Larger Runners vs Managed Runners: コストと速度

> GitHub larger runners とサードパーティのマネージドランナー: 分単価、課金モデル、速度。GitHub の大型ランナーが有効な場面と、マネージドが勝る場面。

Source: https://latchkey.dev/ja/learn/compare-runners/larger-runners-vs-managed-runners  
Updated: 2026-06-25

GitHub larger runners は「ビルドにもっとコアが必要」問題を解決しますが、最初の1分から課金され、料金はプレミアムです。

標準ランナーでは小さすぎる場合、GitHub は大型ランナー (4〜64 vCPU) を提供します。動作はしますが、無料枠なしで最初の1分から高い料金で課金されます。マネージドランナーは同じサイズをはるかに安く実現します。

## Comparison

|  | GitHub larger runners | Managed (Latchkey) |
| --- | --- | --- |
| 無料分 | なし - 1分目から課金 | 無料枠あり |
| 4 vCPU 分単価 | ~$0.016 | ~$0.005 |
| 8 vCPU 分単価 | ~$0.032 | ~$0.01 |
| 自己修復 | なし | あり |
| セットアップ | runner group の設定 | label の入れ替え |

## larger GitHub runners が適する場面

コアがたまに多く必要なだけで、サードパーティのセットアップをまったくしたくないなら、larger GitHub runners が最もシンプルな道です。ただし高い分単価は覚悟してください。

## マネージドが勝る場面

恒常的または大量の重いビルドでは、マネージドランナーは同じ vCPU を約3分の1の価格で提供し、その上にキャッシュと自己修復を備えます。

## How to evaluate a managed runner honestly

- Compare at equal machine shape. A lower rate on fewer vCPUs is not cheaper per unit of work.
- Check billing granularity: per-minute rounding costs real money across a wide matrix of short jobs.
- Include queue and boot time. Cheaper per minute but slower to start can cost more per merge.
- Count your re-runs. Paying twice for the same work is invisible on every rate card.
- Switching is a one-line `runs-on` change in both directions, so a two-week trial beats modelling.

## 結論

たまの大型ビルドなら larger GitHub runners で十分です。恒常的な重いビルドなら、Latchkey のようなマネージドランナーは同じコンピュートを圧倒的に安く提供します。

## FAQ

### GitHub Larger Runners vs Managed Runners: Cost & Speed?

When standard runners are too small, GitHub offers larger runners (4–64 vCPU). They work, but they bill from the first minute with no free allowance and at a high rate. Managed runners hit the same sizes for much less.

### When larger GitHub runners make sense?

If you need more cores only occasionally and want zero third-party setup, larger GitHub runners are the simplest path - just expect a steep per-minute rate.

### When managed wins?

For sustained or high-volume heavy builds, managed runners deliver the same vCPU at roughly a third of the price, with caching and self-healing on top.

### Which should I choose?

Occasional big builds: larger GitHub runners are fine. Regular heavy builds: managed runners like Latchkey are dramatically cheaper for the same compute.

---

Latchkey runs CI/CD that repairs its own failures. Agent entry points: https://latchkey.dev/agent.txt, https://latchkey.dev/openapi.json, https://latchkey.dev/llms.txt