Google Cloud instance

Run GitHub Actions on Google Cloud a3-highgpu-8g

Google Cloud a3-highgpu-8g — 208 vCPU, 1872 GiB RAM, x86_64, NVIDIA H100 80GB. 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.

Get started free Docs

Why this fits

208 vCPU and 1872 GiB RAM — sized for workloads the default GitHub-hosted runners can't comfortably host, all on your own Google Cloud account.
H100 × 8 — Hopper-class CI for foundation-model teams.
NVIDIA H100 80GB for PyTorch / JAX and other CUDA workloads, with drivers + CUDA pre-baked in the runner image.

.cirun.yml

1runners:
2  - name: gcp-a3-highgpu-8g
3    cloud: gcp
4    instance_type: a3-highgpu-8g
5    # Use a GCP Deep Learning VM image family for pre-installed CUDA +
6    # drivers.
7    machine_image: projects/deeplearning-platform-release/global/images/family/pytorch-latest-cu121
8    labels:
9      - cirun-gcp-a3-highgpu-8g

Drop this in your repo root. The first workflow that requests the runner label spins this configuration up on your cloud account.

More runners on Google Cloud

NVIDIA H100 80GB

GPU model details

Google Cloud n2-standard-8

8 vCPU · 32 GiB

Google Cloud c3-standard-8

8 vCPU · 32 GiB

Google Cloud c3-standard-22

22 vCPU · 88 GiB

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.

Get started free Docs

Sources · verified 2026-05-25