Roboflow + Google Cloud VM
Run Roboflow models and Workflows on a Compute Engine VM in your Google Cloud project, provisioned with one Inference CLI command.
How to connect Google Cloud VM to Roboflow
5 stepsInstall the CLI and SkyPilot
On your own machine, run `pip install inference "skypilot[gcp]"`.
Authenticate with Google Cloud
Configure the gcloud CLI for your project and follow SkyPilot's cloud account setup.
Deploy the server
Run `inference cloud deploy --provider gcp --compute-type gpu`, or `--compute-type cpu` for a CPU-only VM. The command prints the deployment name and the server address on port 9001.
Point your client at the server
Set `api_url` to the VM's address on port 9001 and run a model or Workflow with your Roboflow API key.
Stop or remove the VM
`inference cloud stop <name>` pauses the VM so you pay only for disk; `inference cloud undeploy <name>` deletes it.
What teams build with Google Cloud VM
Keep images inside your Google Cloud project
Run inference in your own project and region when images cannot leave your infrastructure.
Run a dedicated GPU for heavy models
Put models such as SAM2 on a GPU VM, where the CPU image would be too slow for practical use.
Prototype on a CPU VM first
Start with a cheaper CPU VM for still images or occasional video frames, then redeploy on GPU when volume grows.
Questions
No. You can create a Compute Engine VM yourself, install Docker and run the roboflow/roboflow-inference-server-cpu or -gpu image as described in the Linux install guide.
No. Inference on your own VM does not use credits; you pay Google Cloud for the machine.
The deploy command opens port 9001 automatically. Check this with your security admin and review Roboflow's guide to securing a self-hosted server.
Why are companies choosing Roboflow?
The end-to-end platform for vision AI: create datasets, train models and deploy to production.
Flexible deployment modes
Move from a scalable cloud API to edge hardware like NVIDIA Jetson without managing complex dependencies.
Integrated workflow builder
Combine multiple AI models, custom business logic and third-party integrations into one production-ready pipeline in a low-code editor.
Model monitoring and alerts
Dashboards track confidence, inference time and request volume for your deployed models, and email alerts tell your team when confidence suddenly drops.
