A fresh VM for every job
Each job boots its own Firecracker VM, and the VM is gone when the job ends. Only the repository's cache disk carries over.
Change one line in your workflow. The job runs on Lizard in its own VM, billed by the second.
1jobs:2build:3-runs-on: ubuntu-latest4+runs-on: lizard-medium-ubuntu-24045steps:6-- uses: actions/checkout@v47+- uses: lizard-build/checkout@v18- run: npm ci && npm test
List price per minute for Linux x64 runners, checked on .
$18.90/ month
This counts toward the usage included with Pro. Pro is $19/month; this is not your total bill.
Pro's $19 of usage covers about 35,000 minutes a month on Medium.
| Provider | 2 vCPU | 4 vCPU | 8 vCPU |
|---|---|---|---|
| Lizard | $0.000274 GB | $0.000548 GB | $0.0010816 GB |
| Namespace | $0.0034 GB | $0.0068 GB | $0.01216 GB |
| Blacksmith | $0.0048 GB | $0.00816 GB | $0.01632 GB |
| GitHub | $0.0068 GB | $0.01216 GB | $0.02232 GB |
| Depot | $0.0068 GB | $0.01216 GB | $0.02432 GB |
Each cell is the price of one minute, with the runner's memory under it. Lizard and Namespace give 2 GB of memory per vCPU; GitHub, Blacksmith and Depot give 4 GB. GitHub's 2 vCPU price is its standard runner for private repositories.
Lizard bills by the second. GitHub rounds each job up to a whole minute, and Namespace bills at least one minute.
Plan fees and free minutes are left out. GitHub Free and Team include $12 and $18 of runner usage a month, Blacksmith gives 3,000, Depot's $20 plan includes 2,000, and Namespace's $100 Team plan lowers its rate by a third. Lizard Pro is $19 a month with $19 of usage included.
Where "a tenth of the price" comes from: the image build job of our own frontend repository. On GitHub's ubuntu-latest runner (2 vCPU, 8 GB) it took a median 3 min 35 s over 54 runs from 21 to 28 October 2026, billed as 4 minutes: $0.024. On Lizard Medium (4 vCPU, 8 GB) the runner VM was up a median 4 min 27 s over 60 runs on 29 October 2026: $0.0024.
A Firecracker VM per job, with a disk that keeps your caches between runs.
Each job boots its own Firecracker VM, and the VM is gone when the job ends. Only the repository's cache disk carries over.
Our Docker actions keep the layer cache on the cache disk, so the next build skips every layer that did not change.
lizard-build/checkout keeps a mirror of the repository on the cache disk. Each job pulls just what changed.
lizard-build/cache stores entries on the same disk, so there is no upload or download. Up to 10 GB per repository.
Jobs from forks start on an empty disk. Branches and pull requests can read the cache; only the default branch writes to it.
Run history, step logs and the VM's console live in the Lizard dashboard, next to your apps and Sandboxes.
The migration wizard reads your workflows and opens the pull request. Or change runs-on by hand.
Step 1
Install the Lizard GitHub App on an organization you admin. Personal accounts aren't supported yet.
Step 2
The wizard suggests a size for every job. Windows and macOS jobs stay on GitHub.
Step 3
Lizard opens one pull request with the changes. Merge it, and the next run starts on Lizard.
Run GitHub Actions on Lizard runners
.github/workflows/build.yml+5−51jobs:2build:3-runs-on: ubuntu-latest4+runs-on: lizard-medium-ubuntu-24045steps:6-- uses: actions/checkout@v47+- uses: lizard-build/checkout@v18-- uses: actions/cache@v49+- uses: lizard-build/cache@v110-- uses: docker/setup-buildx-action@v311+- uses: lizard-build/setup-docker-builder@v112-- uses: docker/build-push-action@v613+- uses: lizard-build/build-push-action@v1
runs-on is optional. Off Lizard, our actions run the stock action with the same inputs and outputs.Lizard runs the job: the VM, its disks and the GitHub runner inside it. Your code, pull requests, workflow files and the Actions tab stay on GitHub.
Change runs-on to a Lizard label, such as lizard-medium-ubuntu-2404. The migration wizard can do this for every workflow and open the pull request for you.
Linux x64 on Ubuntu 24.04, in three sizes: 2, 4 and 8 vCPU. Windows, macOS and ARM jobs stay on GitHub for now.
By the second, at Sandbox prices: Medium (4 vCPU, 8 GB) costs $0.0324 an hour. Runner time comes out of the $19 of usage included with Lizard Pro each month; past that it is pay as you go. The Lizard account that connects the organization pays.
Not yet. Runners work for GitHub organizations, and only an organization admin can connect one.
Yes. Lizard keeps a cache disk only for private repositories, and jobs from forks never run with it. Pull requests and other branches can read it, but only pushes, schedules and manual runs on the default branch save to it.
Yes. Off Lizard they run the stock action with the same inputs and outputs, so moving a job back to GitHub is the same one-line change.
Runners are in private beta. Start a trial, then use Talk to the founder to tell us your GitHub organization, and we'll turn runners on for your account.
You don't need a platform team to go live. Your whole cloud, one CLI command away.