BreakPoint Drop

Blog · Licensing

How to choose an open source license

MIT, Apache 2.0, BSD, GPL, AGPL, BSL, SSPL, Elastic — which license should you pick? The 2026 decision tree.

Published July 13, 2026 · ~6 min read

The 80% answer

MIT or Apache 2.0. MIT is the most permissive (anyone can do anything with the code). Apache 2.0 is similar but adds an explicit patent grant. For most projects, pick one of these and move on. The other licenses (GPL, AGPL, BSD, MPL) are for specific use cases.

The 2026 license landscape

There are three categories of license to consider:

1. Permissive (MIT, Apache 2.0, BSD)

Anyone can do anything with the code, including using it in proprietary software. The most popular licenses in modern OSS. Apache 2.0 is slightly more "corporate" — it includes a patent grant, which is important for companies that want to use the code without patent risk. MIT is the most familiar and the most universally accepted. Both are safe for adoption.

2. Copyleft (GPL, LGPL, AGPL, MPL)

Anyone who distributes a derivative work must also distribute the source under the same license. GPL is the original copyleft license — it's why Linux is open. LGPL is a weaker version for libraries. AGPL is the strongest — it requires the source of any network service that uses the code to be published, which is why some companies refuse to touch AGPL-licensed code. MPL is file-level copyleft — modifications to MPL files must be shared, but the rest of the project can be proprietary.

3. Source-available (BSL, SSPL, Elastic)

NOT open source by the OSI definition. You can read the code but commercial use is restricted. BSL (Business Source License) is used by HashiCorp — converts to Apache 2.0 after a few years. SSPL (Server Side Public License) is MongoDB's license — requires the source of any service that uses the code to be published. Elastic License is Elastic's — restricts hosting by cloud providers. If you want true open source, use MIT/Apache/GPL. If you want source-available, use BSL/SSPL/Elastic — but be aware forking is harder.

The 2026 decision tree

  1. Want maximum adoption? MIT.
  2. Want patent safety for corporate users? Apache 2.0.
  3. Want to ensure derivatives stay open? GPL (libraries → LGPL, network services → AGPL).
  4. Want to monetize the cloud version while keeping the code source-available? BSL with a future Apache 2.0 conversion.
  5. Want to prevent cloud providers from monetizing your work? AGPL (or move to BSL/SSPL like MongoDB and Elastic did).

The maintainer-handoff angle

The license you pick affects how easy the handoff is. MIT and Apache 2.0 are the easiest to fork — anyone can take the code, change the name, and ship a derivative. GPL is harder — the derivative must also be GPL, which limits who can adopt. AGPL is hardest — most companies won't touch it. BSL and SSPL are even harder — the new maintainer may have to negotiate a license change with the original maintainer or the foundation.

If you want the handoff to be frictionless, pick MIT or Apache 2.0. The 41% revival rate from the 2019 arXiv study assumes a permissive license. With a copyleft or source-available license, the rate is much lower.

Frequently asked questions

What license should I use for a new project?

MIT or Apache 2.0. For most projects, pick one and move on.

What about GPL vs MIT?

MIT is permissive (anyone can use in proprietary). GPL is copyleft (derivatives must also be GPL). AGPL is the strongest.

What about source-available licenses?

Not OSI-open-source. BSL, SSPL, Elastic all restrict commercial use. Forking is harder.

Does the license matter for adoption?

Yes. AGPL, SSPL, BSL are deal-breakers for many companies. MIT/Apache/BSD/GPL are safe.

Related reading

Companion

Open source vs source available — what's the difference?

The BSL/SSPL/Elastic deep dive.

Get rescued projects in your inbox

A short Monday-morning note on the projects abandoned last week and who adopted them. No spam, unsubscribe anytime.

Related reading on BreakPoint

The concepts, guides, and reports behind this post.

Concept

Bus factor

The definition, with data and citations.

Guide

How to hand off any project

Step-by-step, with the verify command.

Glossary

Glossary: open source

The term, with citation to upstream sources.