OpenRoots

ORA-2.2

An AI model provider deploying it for a paying client, under ORA 2.2

Trains, fine-tunes, or serves machine learning models commercially. Installing or operating the Work on infrastructure the customer owns and controls. This resolves to the Canopy tier.

Tier

Canopy

Owed

0.5% of revenue attributable to products or services that depend on the Work, capped at $250,000 per year, plus a quarterly Canopy Report.

Sections in play

1.8, 1.9, 3.1, 5.1, 5.2

Who and what

The situation being resolved.

Actor

An AI model provider

Trains, fine-tunes, or serves machine learning models commercially.

over 2M USD

Situation

Deploying it for a paying client

Installing or operating the Work on infrastructure the customer owns and controls.

Rulings

3 questions resolved against the text.

clear

Which tier applies?

Canopy. Trains, fine-tunes, or serves machine learning models commercially. Section 5.2 reporting applies; the Section 5.1 royalty attaches when the Licensee makes Production Use and only to attributable revenue.

Sections 1.8, 3.1, 5.1

clear

Is anything owed?

0.5% of revenue attributable to products or services that depend on the Work, capped at $250,000 per year, plus a quarterly Canopy Report.

Sections 5.1, 5.2

clear

Does this count as Production Use?

Yes. Installing or operating the Work on infrastructure the customer owns and controls. Section 1.9 covers internal operation at operating scope and systems made available externally to customers or the public. Development, testing, evaluation, research, and personal non-revenue use remain outside it.

Sections 1.9

Cautions

What this situation gets wrong most often.

  • Deploying on a customer's own infrastructure is carved out of Section 4 by 4.2, but the Section 5 royalty still attaches to revenue you earn from work that depends on the Work.

  • Attribution lives in the package manifest under ORA 8.1, not only in documentation, because an agent package is installed by a machine that never reads a directory listing.

Same actor

Other situations for this party.

Evaluating it internallyOperating it internallyShipping it in productionEmbedding it in a product you sellOffering it as a hosted serviceForking and modifying itRedistributing it without chargeReselling itRepackaging itPorting itListing it in a marketplaceUsing it in a larger productDeploying to customer-controlled infrastructureTraining a model on itIncluding it in a datasetUsing it for retrieval augmentationCrossing the revenue thresholdBeing acquiredChecking licence continuityCombining it with other licencesNaming the project publiclyContributing backBeing auditedDiscovering you are in breachThe Licensor disappearsProducing an SBOM

Non-normative. Where this and the licence text disagree, the licence text applies.