Get Started
Open source that pays

Mission 6 of 6

Pick the money model that fits your project.

Stars don't pay rent. Compare six honest ways open source earns, choose one from your own evidence, and get its first three steps done today.

60 minIntermediateFree to set upStep 1 free

What you'll do

Compare the six money models

See who pays, what licence each one needs, and the main risk, side by side.

Decide from your own evidence

Use your issues, users and licence to choose one model, and write the decision down.

Set up your first three steps

Turn on Sponsors, wire a Pro tier for the worked example, and start the model you picked.

  • A written decision you can reread when revenue is slow
  • A licence check that fails gracefully
Before you start
  • Your project from this track, with at least a few weeks of issues, Discussions and office hours behind it. The decision needs real signal about who uses it and what they ask for
  • The licence you chose in mission 1, because it rules some models in and out
  • A way to take money: a Stripe account for direct billing, or a merchant-of-record account (Polar, Lemon Squeezy, Paddle). Take payments with Stripe covers the Stripe side
Your progress0/5
01/05The six models, with the trade-offs nobody puts on the pricing page0 done
STEP 01 of 05~12 min

The six models, with the trade-offs nobody puts on the pricing page

See all the options side by side before you fall for the first one, because each one changes what the project becomes.

Before you decide how your project earns, look at the six options and what each one costs you, including the trade-offs that never make it onto a pricing page.

  1. Read the table below. It shows, for each of the six models, who pays, what licence it needs, the realistic ceiling, your hours and the main risk.

    The models
    Model                 Who pays              Needs licence?   Typical ceiling      Your hours     Main risk
    Sponsors / OpenColl.  individuals, a few    no               pocket money to      near zero      nobody pays for
                          companies                              part-time income                    what they get free
    Paid support /        companies using it    no               day-rate income      sells hours    you become a
    consulting            at work                                                                    contractor
    Pro tier / add-on     individuals + small   no               a real side income   build + bill   the free tier
                          teams                                  if the job matters                  is too good
    Hosted / open core    teams who won't       MIT ok, AGPL     a business           most of them   you now run a
                          self-host             helps                                                SaaS, with uptime
    Commercial licence    companies, legal      AGPL/BSL         per-deal, lumpy      sales calls    long cycles, CLA
                          dept involved         required                                             needed day one
    Bounties              one company at a time no               per feature          build on spec  features you'd
                                                                                                     have said no to

    Read the 'Your hours' and 'Main risk' columns first. The money columns are what everyone looks at; the other two are what decides whether you still like the project in a year.

    Now the detail behind each row, one model at a time.

Sponsors and Open Collective

Turning on GitHub Sponsors takes ten minutes and you should do it today regardless of anything else, because it's free and the button has to exist before anyone can click it. But understand what it is: a tip jar.

Donations follow gratitude, and gratitude is a weak, slow signal. The maintainers it works for have either a very large user base, or a few companies sponsoring because the tool is in their critical path and a sponsor tier includes a logo in the README and priority on issues.

If you go this route, make tiers that give companies something to show their boss: the logo, a 'sponsors get a reply within 2 working days' promise, a named contact.

Open Collective is the same idea with transparent accounting and the ability to pay out to multiple contributors, which matters once you have a second maintainer.

Paid support and consulting

This is the fastest money, because a company that uses your tool in production will pay for a day of your time to integrate it, or a retainer to be allowed to email you. No licence needed, no product to build.

The trap is that it sells hours, so it doesn't scale, and the consulting work pulls you into their problems and away from the project. Good as a bridge, dangerous as a destination.

A clean version: a fixed-price 'integration day' and a monthly 'priority support' retainer, both on a page, both with a scope line.

A Pro tier or paid add-on

The free tool stays complete for the job it does. The paid part does a different, adjacent job that the people with budget need: team features, reporting, integrations, a GUI, a hosted dashboard, compliance output. Individuals never hit a paywall for the thing the README promised.

This is the model that fits most CLI tools and libraries, needs no server if you're careful, and bills through a merchant of record so you never think about VAT.

The risk is the line between free and paid. If the free tier is so good nobody needs Pro, no revenue. If you move a free feature behind the paywall, you lose the community and the stars in a week and HN will write about it.

A hosted version (open core)

You run it as a service for people who don't want to self-host. The free tool is the on-ramp; the hosted version is where the money is.

It works with MIT, and works better with AGPL. A cloud giant can't take your MIT code and offer it hosted without competing on even terms; under AGPL they must open source their version, which they won't.

This is the biggest ceiling and the biggest change: you now run a SaaS, with uptime, support, security and billing. It's a business, not a side project. The whole Launch a SaaS track applies.

Commercial licensing (dual licence)

This is only possible if you chose AGPL or a source-available licence in mission 1 and had a CLA from day one. Companies whose lawyers won't accept AGPL pay for a commercial licence to the same code.

Deals are lumpy, take months, and need you to answer procurement questionnaires. It's a real model (databases and embedded libraries live on it) but it's a sales job, and a solo maintainer with an MIT project can't start it now without relicensing, which needs every contributor's consent.

Bounties

A company pays for a specific feature, via a bounty platform or direct. Fine as occasional income; bad as a model, because it pulls the roadmap toward whoever pays, which is how you end up building the thing you said no to in mission 5.

Accept bounties only for features already labelled feature, never to decide what's in scope.

Warning
The one line not to cross, under any model: never take away something that was free. Not a feature, not a flag, not a platform. Add paid things next to the free thing. Projects that moved free features behind a paywall are the cautionary tales people cite on every open source thread, and the forks that replaced them are named after the betrayal.
Checkpoint
For each of the six models you can say, in one line, whether your licence allows it and who would pay.
If it breaksI'm on MIT and wish I'd picked AGPL
I'm on MIT and wish I'd picked AGPL
If you're the only contributor so far, you can still change it: update LICENSE, note it in the CHANGELOG, and accept that anyone who already has the MIT version keeps those rights. If others have contributed, you'd need each one's written consent for their code. Usually it's easier to keep MIT and pick a model that doesn't need the licence: Pro tier, hosted, support.
Still stuck? Ask the community
Next mission · Start an AI agencyPick a niche and an offer businesses already pay for (and stop calling yourself an AI agency)
→