There is no hoster share in License v2.0
Earlier license versions reserved a hoster configuration field that was never operative. License v2.0 (Direct Settlement) removes it entirely: no hoster allocation, no reserved field, no hoster payment path. Every royalty dollar flows through each model's published split — most of it to the original creators. Hosters are very welcome at OMLA, just not as a royalty line.
What changed in v2.0
- The reserved hoster field is gone. Versions 1.0–1.2 stored a 0–20% hoster configuration that was never operative — any non-zero value simply blocked processing. v2.0 removes the field from the license and the registry instead of carrying dead scaffolding forward.
- No entitlement, no expectation. License v2.0 §6 states it plainly: no hoster allocation, no hoster payment path, and nothing in this version promises one. A future hoster allocation would require a later license version, prospectively accepted by creators.
- The economics went to the lineage instead. Money now flows recursively through each model's published split — a fine-tune retains at most 5%, and the rest flows upstream to the original creators. See the License for the full rules.
Two ways hosters participate today
As a publisher
If you train, fine-tune, or merge models you have the rights to license, register them like any creator: sign the manifest, publish your split and payment pointers, and commercial users pay your wallets directly. Start at Publish or read the creator guide.
As a platform
Integrate the signed registry and ship the open omla-resolve algorithm so every model you serve shows its license terms and where the money goes. The integration surfaces are small and read-only — see For Platforms.
One more hat to be aware of: if you monetize hosted inference of a registered model — selling API access, for example — you are a commercial user under the License: self-assess the 30% royalty, resolve it with the open algorithm, and pay the creators' wallets directly. See the Users page.
Questions
Hosting-provider questions and integration help: hello@omla-ai.org.