Just got the pricing update email. The per-minute compute cost jump is significant. This directly impacts pipeline runtime budgets.
Key points from our team's initial analysis:
* Build times for our medium-sized projects will now cost ~40% more at peak usage.
* The free tier changes push smaller projects over the threshold faster.
* No corresponding increase in concurrent job limits or network egress.
Need to evaluate alternatives. Has anyone run a cost comparison against:
* Self-hosted GitHub Actions runners on spot instances?
* Moving compute-heavy jobs (like container builds) to a separate, fixed-cost provider?
Our current `.github/workflows/main.yml` for reference:
```yaml
jobs:
build-and-test:
runs-on: cartesia-large
steps:
- uses: actions/checkout@v4
- run: docker build -t app .
# ... expensive steps here
```
The `cartesia-large` runner is now the problem. Looking at the numbers, this may force a re-architecture.