My open-source alternatives directory's Airtable page lists four repositories. The numbers here are from the 2026-10-18 refresh. Stars, forks, language and last push come from saas.json. Releases and open items come from evidence.json, generated at 06:03 UTC that day.
The headline isn't a number. GitHub's API reports the license of all four repos as NOASSERTION, meaning it couldn't match the LICENSE file to a single SPDX identifier. So I read the files.
The table
"Open items" is GitHub's open issue count, which also includes open pull requests.
| Repo | Stars | Forks | Language | Latest release | Releases, last 12 mo | Open items |
|---|---|---|---|---|---|---|
| nocodb/nocodb | 65,294 | 5,095 | TypeScript |
2026.09.1, 2026-09-29 |
23 | 713 |
| directus/directus | 38,571 | 4,977 | TypeScript |
v12.5.0, 2026-10-07 |
30 | 367 |
| teableio/teable | 21,896 | 1,357 | TypeScript | 2026-10-16 | 100+ | 146 |
| baserow/baserow | 6,169 | 767 | Python |
2.4.1, 2026-10-13 |
24 | 1,175 |
None are archived. All four were pushed between October 16 and 18.
What the LICENSE files say
These are from my directory's manual override file, each checked against a pinned commit on 2026-09-05:
-
NocoDB: Sustainable Use License 1.0. Its
LICENSE.mdsays the master and develop branches are under that license and other branches are not licensed. The Sustainable Use License is not an OSI-approved open-source license. - Directus: Monospace Sustainable Core License 1.0 (MSCL-1.0-GPL). The file's Permitted Purpose excludes Competing Use, and it states a later conversion to GPL-3.0.
-
Teable: AGPL-3.0 for the core apps (
apps/nestjs-backend,apps/nextjs-app), with Section 7 brand-protection terms, and MIT forpackages/. -
Baserow: MIT for the Baserow Open Source Edition.
premium/andenterprise/have their own licenses, anddocs/is CC BY-SA 4.0.
So the most-starred repo in this category and the second one both use source-available licenses with use restrictions. If "open source" in your search means OSI terms, Teable's core and Baserow's open edition are the two that qualify, each with a carve-out to read. That's the same pattern I found comparing password managers by their LICENSE files, and it's why the directory can't print some GitHub license values verbatim.
Releases: one number is a cap
Teable's "100+" isn't a count. My ETL reads one page of 100 releases from GitHub, and when all 100 fall inside the last year it marks the count as capped (releases_last_year_capped: true). Its latest tag is release.2026-10-16T07-06-34Z.3445, which looks like a timestamp from automated release tooling, not a hand-picked version number. A count like that tells you how Teable tags releases, not how much it changes per release. I wrote about why release counts measure tagging habits on a broader set of repos.
The other three sit close together: Directus at 30, Baserow at 24, NocoDB at 23. All three tagged a release within the three weeks before the refresh.
Two ratios worth noticing
Open items per star. Baserow has the fewest stars here (6,169) and the most open items (1,175). That's about 0.19 open items per star, against roughly 0.01 for NocoDB and Directus. I'm not going to read that as quality. A project that takes its issue tracker seriously on GitHub can have more open items than one that triages elsewhere. But if you judge Baserow by stars alone, you're probably underrating how much activity the repo has.
Stars per fork. NocoDB is about 12.8, Directus about 7.7, Baserow about 8.0, and Teable about 16.1. Teable's higher ratio means fewer people fork it relative to how many star it. I don't have data on why.
How I'd read it
NocoDB has by far the biggest following and a regular release cadence, under a license that restricts use. Directus ships most often of the three with conventional version numbers, and the directory's editorial notes say it fits better as a headless CMS backend than as a spreadsheet replacement. Teable ships constantly under AGPL. Baserow has an MIT open edition and a large open backlog.
None of these numbers measure Airtable features. The directory's notes say none of the four replicate Airtable Automations' marketplace or its interface designer. That's an editorial judgment, not a measurement, so test it with a trial install.
I left out a "What real users say" section. I couldn't collect a comment sample for these four this run, and I won't write one without counted, linkable sources.
Part of an ongoing 6-month experiment running three AI-curated directory sites. The technical claims here are real; this article was AI-assisted.
Top comments (0)