Google Cloud instance
Run GitHub Actions on Google Cloud a2-highgpu-1g
Google Cloud a2-highgpu-1g — 12 vCPU, 85 GiB RAM, x86_64, NVIDIA A100 40GB. Cirun spins one up per workflow on your Google Cloud account and tears it down when the job finishes — you pay Google Cloud for the compute, nothing per CI minute.
Why this fits
- 12 vCPU and 85 GiB RAM — sized for workloads the default GitHub-hosted runners can't comfortably host, all on your own Google Cloud account.
- Single A100 — ML eval suites and small fine-tunes.
- NVIDIA A100 40GB for PyTorch / JAX and other CUDA workloads, with drivers + CUDA pre-baked in the runner image.
.cirun.yml
.cirun.yml
'/%3e%3cpolygon%20fill='%23231F20'%20fill-opacity='.7'%20points='0%200%20.936%200%202.5%202%203.979%200%205%200%205%205%204%205%204%201.7%202.5%203.5%201%201.7%201%205%200%205'%20transform='translate(6%2010)'/%3e%3cpolygon%20fill='%23231F20'%20fill-opacity='.7'%20points='0%200%201%200%201%204%203.5%204%203.5%205%200%205'%20transform='translate(12%2010)'/%3e%3c/g%3e%3c/svg%3e)
1runners:2 - name: gcp-a2-highgpu-1g3 cloud: gcp4 instance_type: a2-highgpu-1g5 # Use a GCP Deep Learning VM image family6 # (deeplearning-platform-release) for pre-installed CUDA + drivers.7 machine_image: projects/deeplearning-platform-release/global/images/family/pytorch-latest-cu1218 labels:9 - cirun-gcp-a2-highgpu-1g
Drop this in your repo root. The first workflow that requests the runner label spins this configuration up on your cloud account.
Ready to run your CI here?
Cirun is free for open source. For private repos, flat monthly plans by repo count — never per CI minute.
Sources · verified 2026-05-25