Pick the cloud per workflow.
Each label runs in one cloud. Out of spot capacity, the next VM type launches, never the next cloud.
Learn morestrategy:
matrix:
cloud: [aws, gcp, azure]
runs-on: backmesh-${{ matrix.cloud }}-4Run GitHub Actions on spot VMs in AWS, GCP and Azure, from one install.
Runners and their autoscaler run in your own accounts, with no vendor access, at up to 20× less compute cost than GitHub-hosted runners.
Runs in your own accounts on
- AWS
- Google Cloud
- Azure
150,000 minutes a month: $1,800 on GitHub, $260–410 on backmesh.
4-vCPU Linux jobs on AWS, all backmesh VMs included; the range runs from evenly spread jobs to bursty ones.
GitHub-hosted$12.00per 1,000 minutes
backmesh on AWS11–12× less$0.95–$1.02 per 1,000 minutes
backmesh on Azure18–20× less$0.59–$0.65 per 1,000 minutes
backmesh on GCP5–7× less$1.62–$2.14 per 1,000 minutes
Pay for jobs started, not minutes.
Each tier caps the jobs started over a rolling 30 days. Going over only warns; an unpaid bill blocks upgrades, never jobs. A subscription links your GitHub orgs through GitHub sign-in, with no licence keys, and installs count jobs with one event per job started (what's sent).
- FreeUp to 5,000 jobs$0Install
- TeamUp to 50,000 jobs$350/yearor $40 monthly. Subscriptions open with customer installs.
- BusinessUp to 200,000 jobs$1,050/yearor $125 monthly. Subscriptions open with customer installs.
- ScaleUp to 500,000 jobs$2,100/yearor $250 monthly. Subscriptions open with customer installs.
- Enterprise500,000+ jobs$4,200/yearor $500 monthly. Subscriptions open with customer installs.
An org links to one subscription; an enterprise link covers its orgs. Tiers change or cancel in the customer portal. See the terms.
One status for every cloud.
Pools, queue, 24 h and 7 d job times, and month-to-date cost, in one command.
Learn more$ backmesh status bm-aws-spot: running, 2 VMs (1–12) bm-gcp-spot: running, 1 VMs (1–2) bm-azure-spot: running, 1 VMs (1–8) backmesh-aws-4: 6 assigned, 6 running 24 h: 42 jobs; queued to started p50 21 s, p95 1.6 min Cost since 2026-11-01: $41.80
Every job log ends with what it used.
Peak memory, OOM kills, CPU and disk, to size each job.
Learn morebackmesh: resources over 312 s,
sampled every 5 s
Memory peak 9.8 GiB of 15.5 GiB,
0 OOM kills
CPU average 2.41 cores,
peak 3.91 of 3.98
Disk work directory peak 3.4 GiBA GitHub App that can't read your code.
Four permissions, its webhook off, and its key held only by the listener.
Learn moreA stolen key can
- Register a runner and take backmesh jobs, with their secrets
- Dispatch or rerun a workflow
- Read, cancel or delete runs and logs
- Write check runs
Through GitHub's API, it can't
- Read your code
- Manage repository secrets
- Change administration settings
- See org members
Before you install
Will it work?Linux x86-64 only. Private repos only.
- No macOS, Windows, Arm or GPU runners.
- Public repositories can't use the runners.
- A reclaimed spot VM fails its jobs; one check run per workflow run lists them to rerun.
- Each job starts with an empty Docker layer cache.
Is it safe?No vendor access. Jobs from every repo share VMs.
- Runners, the listener and the API run in your accounts.
- Up to 8 jobs share a VM, in privileged containers.
- Run code you trust as you trust your org's members.
- Per-repo isolation is planned.
What does it cost?An always-on VM per cloud, until scale-to-zero ships.
- Your clouds bill you directly.
- Before any job runs: about $245–260 a month for AWS, $635–750 for three clouds.
- After scale-to-zero: about $80 a month on AWS.