The association behind the license
OMLA — the Open Model Licensing Association — is organizing as a Washington non-profit with one job: make it normal for the people who create open-weight AI models to share in the money those models earn. It maintains the OMLA Public License, runs the public registry of signed model manifests, and publishes the open resolver — then gets out of the way while commercial users pay creators directly. It keeps no records of usage, payers, or payments, because it never receives them.
Built by tinkerers who kept finding their work inside other people's products
OMLA started as a personal project by its founder, a builder who trains and tunes models on his own hardware, after watching a pattern repeat across the open-model ecosystem: someone pours months into a model, releases it openly, and then finds it quietly powering commercial products that return nothing to anyone who made it. The usual answers were bad ones: lock everything behind proprietary licenses, or accept that open means unpaid.
The OMLA Public License is the third answer. Non-commercial use — research, learning, hobby work — stays completely free, no strings attached. Commercial use above a revenue floor gets a choice: release the derivative's own weights back under the same license, or pay 30% of applicable revenue directly to the creators — and most of that money flows to the original creators up the lineage. The association exists to make that choice practical: a registry that proves whose work a model is, a published split, and arithmetic anyone can check and run themselves.
Four commitments that shape everything
No custody, ever
OMLA never holds, moves, or even sees money. It publishes the split; commercial users compute what they owe and pay creators directly.
Creators first
Most of the money flows to the original creators. The suggested defaults pass 95% of a fine-tune's royalty upstream through the lineage — and whatever split a creator declares is public and signed. OMLA takes no cut of any royalty.
Your choice, every quarter
Copyleft or pay — release your derivative under the same license, or pay the people whose work it's built on. No lock-in, no telemetry, no third option forced on anyone.
Honest in public
The license says plainly what it governs, who owes what, and how to comply. What isn't built or finalized yet says so — not buried in a FAQ.
A young non-profit, run in the open
OMLA is organizing as a Washington non-profit; incorporation and federal 501(c) filings are in progress, so donations are not yet tax-deductible. It charges nothing on royalties by design — the license, the registry, and the resolver are the entire product, and none of them take a cut.
Governance sits with a founding board — currently one founding member and two open seats being actively recruited, alongside calls for legal and technical contributors to help finalize the license and the tools around it. Decisions that affect the license and the registry are deliberate and human.
Meet, join, ask, or help fund it
Board & governance
Who steers OMLA, the founding member's background, and the two open board seats.
Press & outreach
The one-sentence description and the press contact.
Contact
One public inbox, read by people — plus the founder directly.
AI disclosure
How AI assistance was used to build this project, what humans decided, and why we say so.
Where to check our work
The license says plainly, on its own page, what it does and who gets paid — effective 2026-08-11. The Terms of Service and Privacy Policy describe exactly what this site does and does not collect. And every published model's public record — identity, lineage, split, and payment pointers — is inspectable in the registry. The plan below says what's built, what isn't, and what happens next.
What exists, and what comes next
Now — operative
The license (OMLA-PL-2.1, effective 2026-08-11), the signed registry, the resolver, and the browser key and wallet tools are live. Model submissions are manual: a signed manifest, sent by email.
Phase 1 — operative license
Outside counsel reviews and finalizes OMLA-PL-2.1. The operative text gets a pinned SHA-256 digest, and an automated publish endpoint plus a manifest-builder-and-signer tool replace the email step.
Phase 2 — harden & grow
Independent registry mirrors and witnesses (so trust doesn't depend on this one origin), a payment-rail conversion partner, and translated pages restored.
OMLA Board
The OMLA Board stewards license evolution, hears concerns about listings, and guides the association's strategic direction. It is not a court: it publishes corrections, not rulings.
Founding Member
Role: Founding Member
- [Redacted for this archive]
- [Redacted for this archive]
- [Redacted for this archive]
- [Redacted for this archive]
- [Redacted for this archive]
- [Redacted for this archive]
- [Redacted for this archive]
[Redacted for this archive]
Full bio
[Redacted for this archive. The original text described the founder's professional background.]
[Redacted for this archive. The original text described the founder's location and background.]
+
Board seat open
Seeking expertise in AI/ML, legal/IP, finance, or policy. Bring your skepticism; we take notes. Get in touch.
+
Board seat open
Looking for international representation. Especially folks outside the Anglosphere — the license has to work everywhere. Get in touch.
Get Involved
OMLA is a nonprofit in active development. We are looking for people with expertise across several areas to help guide the association and build the platform.
Join the Board
Board members help set policy, steward the license text, and guide strategic direction. Seeking AI/ML, legal, financial, and policy expertise. Email contact@example.org.
Legal & Integrity Officers
Legal counsel to provide guidance on the operative license. Legal expertise in AI licensing, copyleft, and cross-border enforceability. Contact contact@example.org.
Technical Contributors
Open-source contributors to the site, the open resolver, and the registry snapshot pipeline. Security reviewers for the client-side key and wallet tools. Contact contact@example.org.
Payment-Rail Partners
Creators publish pointers on whatever rail they prefer — Lightning, on-chain, Stripe, PayPal — and payers don't always hold funds on the same rail. OMLA will never operate payment infrastructure (the no-custody rule is permanent), so there's real room for a plug-and-play converter between rails. If you build one, we want to talk.
Requests & Disputes
Email contact@example.org with a clear description of the issue, relevant model IDs or wallet addresses, and any supporting documentation. This is an informal, good-faith review — OMLA keeps no payment records and makes no legal determinations (see Terms §6.1 and §7.4) — and corrections it makes to the registry are published for transparency.
Press
The short version, ready to quote.
Elevator description (1 sentence)
OMLA is a public license (OMLA-PL-2.1, operative 2026-08-11) for open AI models with a copyleft-or-pay trigger: free for non-commercial use; commercial use above a $200K revenue floor — or any size directly selling the model's outputs — either releases the derivative's own weights under the same license, or pays 30% of applicable revenue directly to the creators named in the model's signed manifest. Nothing is reported to OMLA, ever.