GPU runner

NVIDIA A10G runners for GitHub Actions

AWS's mid-tier GPU. Cheap enough for per-PR inference smoke tests. Cirun spins NVIDIA A10G VMs up on your cloud account per workflow — pick whichever cloud has the SKU on your existing account, CUDA + drivers come from each cloud's own Deep Learning image.

Run a workflow on NVIDIA A10G Docs

Why this fits

24 GB VRAM and ~125 TFLOPS FP16 — inference smoke tests, fine-tune sanity checks and per-PR GPU regressions.
Works with PyTorch, vLLM, Stable Diffusion — CUDA + drivers come from your cloud's official Deep Learning AMI / image, not your workflow.
Ephemeral by default — each job gets a clean VM, so leaked weights or hung CUDA contexts can't poison the next run.

.cirun.yml

1runners:
2  - name: gpu-a10g
3    cloud: aws
4    instance_type: g5.xlarge
5    # Use AWS Deep Learning AMI GPU PyTorch on Ubuntu 22.04
6    # (Marketplace) for pre-installed CUDA + drivers, or the
7    # Cirun-published NVIDIA AMI.
8    machine_image: ami-04823729c75214919
9    labels:
10      - cirun-gpu-a10g

.cirun.yml

1runners:
2  - name: gpu-a10g
3    cloud: aws
4    instance_type: g5.2xlarge
5    # Use AWS Deep Learning AMI GPU PyTorch on Ubuntu 22.04
6    # (Marketplace) for pre-installed CUDA + drivers, or the
7    # Cirun-published NVIDIA AMI.
8    machine_image: ami-04823729c75214919
9    labels:
10      - cirun-gpu-a10g