MARS
Join waitlist
The cost of more builds

Invest in your own hardware

Job-minutes per month

About 1,000 CI runs per month at an average of 10 minutes each.

Monthly cost · USD
GitHub
$608
Blacksmith
$788
Azure VM
N/A
boat.dev
N/A
MARS
$0

macOS

GitHub M1, 3 vCPU · Blacksmith M4, 6 vCPU

Monthly cost · USD
GitHub
$88
Blacksmith
$68
Azure VM
$30
boat.dev
N/A
MARS
$0

Windows

2 vCPU · x64 runners

Monthly cost · USD
GitHub
$48
Blacksmith
$28
Azure VM
$14
boat.dev
$20
MARS
$0

Ubuntu

2-vCPU x64 Linux runners · boat.dev small VM: 2 vCPU / 4 GB

Rates, allowances and comparison assumptionsMonthly costs in USD. GitHub Free and Blacksmith private-repository allowances included. Rates checked 31 October 2026 at 15:06 GMT.

MARS runtime pricing. Hobby includes 10k self-hosted job-minutes per rolling 28 days. Entrepreneur costs $5/month and includes 70k job-minutes per rolling 28 days. Private-repository allowances only: 10k minutes exceeds Blacksmith’s 3,000 x64 2-vCPU Linux minutes. GitHub Free’s $12 compute credit equals about 194 macOS, 1,200 Windows or 2,000 Ubuntu minutes at rates shown. This does not claim to beat standard GitHub-hosted public-repository jobs, which can be free. Hardware, electricity, maintenance, network, taxes and additional licences are extra. Free-tier availability is subject to capacity.

Durations are equal, but hardware performance may differ. macOS compares GitHub’s standard M1 runner (3 vCPU, $0.062/min) with Blacksmith’s M4 6-vCPU runner ($0.080/min). Windows x64 2-vCPU: $0.010/min vs $0.008/min. Ubuntu x64 2-vCPU: $0.006/min vs $0.004/min. boat.dev uses its small 2-vCPU / 4-GB sandbox at $0.018/hour. It is a general-purpose sandbox, not a dedicated hosted CI runner. Windows on Blacksmith is in public beta.

Each operating system models a separate private-repository workload with the full monthly allowance available: GitHub Free includes $12 in compute usage; Blacksmith includes 3,000 x64 2-vCPU Linux minutes, equivalent to $12. Each hosted estimate is max(0, minutes × rate − $12). Credits are shared across an organization’s workloads, not per operating system: do not add these cards with three credits for one organization. GitHub Pro and Team include $18; Enterprise Cloud includes $300. Standard GitHub-hosted public-repository runners can be free. Storage, caching and other add-ons are excluded.

boat.dev estimate. Uses the small 2-vCPU / 4-GB VM at $0.018/hour. Assumes each billed VM-minute runs at 100% CPU load; boat.dev charges by VM size and running time, not utilization. Plan payments are credited toward VM time, with a $20/month minimum plan; 200,000 job-minutes cost $60 in VM time before plan-tier limits. 17,000 12-minute runs need the $100 plan tier at 840 starts/day, so monthly plan cost is at least $100 for that workload. Start overages, concurrency and VM startup time are not included in the chart; actual cost depends on job count, run duration and required plan tier.

Azure models self-managed GitHub Actions runners on East US Standard_D2as_v5 VMs (2 vCPU, 8 GiB), not Azure Pipelines. Pay-as-you-go compute: Windows Server $0.178/hour, Ubuntu/Linux $0.086/hour. Cost = job-minutes ÷ 60 × hourly rate, with one job per VM and immediate deallocation between jobs. Windows Server licensing is included. Disks, startup/idle time, network, storage, management and taxes are excluded; this is not a total Azure bill. No promotional credits, Spot, reservations or savings plans are assumed. No equivalent Azure macOS VM is modelled.

GitHub runner rates · GitHub Free allowance · Blacksmith runner rates · Blacksmith allowance · Azure retail rates · boat.dev pricing and limits · MARS plans

Simple plans for your fleet

Start free. Grow on your terms.

One subscription for your fleet. Your hardware runs the jobs; MARS hosts the control plane.

For hobbyists

Hobby

For personal projects and your first worker.

Start here
£0 / $0/ month

INCLUDED

Self-hosted workers
3 workers
Parallel job capacity
Up to 5
Ops
500 ops
CI minutes
10k
Full observability
1 day
Aggregation history
7 days
FOR LARGER ORGANISATIONS

Scale

For larger fleets and tailored requirements.

For your requirements
Let’s talk.
Talk to us

LET’S PLAN YOUR

  • Private deployment
  • Self-hosted worker fleet
  • Parallel job capacity
  • Ops, scoped together
  • Full observability retention
  • Aggregation history
  • Single sign-on (SSO)
  • SOC 2 requirements
  • Priority support

Capacity, retention and security, scoped together.

* Unlimited jobs, not unlimited hardware. Actual concurrency depends on your workers’ CPU, memory, disk and configured limits.

Reviews and sandbox actions use Ops. Different actions can use different amounts.

Full observability keeps detailed job metrics. Aggregation history keeps longer-term summaries.

Your workers. Your resources. MARS handles the control plane.Compare the cost of your hardware

Get notified when MARS launches

MARS / PLAN PREVIEW