Complete Run:ai keyboard shortcuts and commands reference — 14 shortcuts across 3 categories. Quick reference cheat sheet for Windows & Mac.
Run:ai schedules GPU workloads on Kubernetes, and the runai CLI is the user-facing side: submit a job with a GPU count, watch it, attach to it, delete it. The commands map onto Kubernetes objects underneath, but the CLI hides pods and namespaces behind projects and jobs. The table follows the current CLI; the notes cover the setup step people skip.
| Shortcut | Action | Description |
|---|---|---|
| runai submit [job] -g 1 | Submit GPU job | Submit a job requesting 1 GPU to the cluster scheduler. |
| runai list jobs | List jobs | Show the current list of jobs and their status. |
| runai describe job [name] | Job details | Show detailed status, events, and resource usage for a job. |
| runai logs [job] | View logs | View a job's stdout/stderr logs. |
| runai delete job [name] | Delete job | Delete a job and free its allocated GPUs. |
| runai attach [job] | Attach to job | Attach your terminal to a running interactive job. |
| Shortcut | Action | Description |
|---|---|---|
| runai login | Login | Authenticate the CLI against the Run:ai control plane via browser SSO. |
| runai whoami | Current user | Show the currently authenticated user and cluster context. |
| runai config project [name] | Set default project | Set the default project so commands don't need -p on every call. |
| runai exec [job] -it bash | Enter job | Open an interactive shell inside a running job container. |
| Shortcut | Action | Description |
|---|---|---|
| runai top node | Node GPU usage | Show per-node GPU utilization across the cluster. |
| runai list projects | List projects | Show projects and their GPU quota allocations. |
| runai delete project [name] | Delete project | Remove a project and its resource quota (admin only). |
| runai nodepool list | List node pools | List node pools available for scheduling, e.g. grouped by GPU type. |
The most essential Run:ai shortcuts are: runai submit [job] -g 1 (Submit GPU job), runai list jobs (List jobs), runai describe job [name] (Job details).
These are command-line commands — type them in your terminal or console. Combine them with shell history search (Ctrl + R) and aliases to work even faster.
The Run:ai shortcut for submit gpu job is runai submit [job] -g 1. Submit a job requesting 1 GPU to the cluster scheduler.
Yes — use My Stack to combine Run:ai shortcuts with any other platform on this site into one printable reference, which is useful if your daily workflow spans several tools.
runai login authenticates through the cluster's identity provider and runai whoami confirms the user. runai list projects shows the projects you may submit to, each with a GPU quota, and runai config project [name] sets the default so later commands need no -p flag. Forgetting this step is the usual cause of "project not found" on the first submit.
runai submit [job] -g 1 submits a job with one GPU; the image, command and fractional GPU counts are further flags, and --interactive keeps the job alive for a notebook or shell. runai list jobs shows status — pending jobs are waiting for quota or nodes — and runai describe job [name] explains why, including scheduler events. runai logs [job] streams output and runai attach [job] connects to the process; runai exec [job] -it bash opens a shell in the job's container for debugging. runai delete job [name] stops it and releases the GPUs.
runai top node shows GPU allocation and utilisation per node, which tells you whether a pending job is blocked by quota or by physical capacity, and runai nodepool list shows the pools an administrator has defined for different GPU types. runai delete project [name] is an administrator action and removes the project's namespace.
Open your assistant with this page preloaded as the source — great for follow-up questions like "which of these work in other apps?"