Skip to content

AI Ownership Index

Documentation for adopting, deploying and operating open-weight AI, drawn from published sources, with the missing parts stated.

New to this? Start here explains the ownership level and the four factors behind it: use and modify, transparency, reliability and data control.

For each open-weight model, and the providers that serve it, the library documents what an organisation needs to decide, in four domains that follow the adopter’s journey.

Assess

Openness, licence, provenance, EU AI Act position, safety, evaluation and the AOI score.

Implement

Installation, hardware, serving, safe-deployment controls, quantisation, fine-tuning and integration.

Use

Capabilities, context window, chat template, languages, tool use and structured output.

Support

Troubleshooting, versions, security disclosure, channels and deprecation.

Each section carries a source badge: aggregated (cited from an existing source), OneHill-tested (produced or verified by OneHill with a recorded method, not yet used) or gap (could not be gathered, with the reason). Every domain shows a completeness meter.

The structure follows Diátaxis, ISO/IEC/IEEE 26514, the Good Docs Project, Model Cards, the Model Openness Framework and the EU AI Act.