Assess
Should we adopt it? Openness, license, provenance, EU AI Act posture, safety, benchmarks, and the AOI trust score.
AI Ownership Index
The complete documentation you need to adopt, deploy, and operate open-source AI - aggregated from what exists, complemented by our own testing, and honest about what can't be provided.
New to this? Start here: what you are actually deciding answers the three questions that matter, in plain language, without benchmarks: can it be restricted, what happens to your data, and will you be locked in.
What this is
Section titled “What this is”A library, not a report. For each open-source model (and the providers that serve it), we document everything an organisation needs to make an informed decision - organised into four domains that follow the adopter’s journey.
Implement
How do we deploy it? Install paths, hardware, serving stacks, quantization, fine-tuning, and safe-deployment controls.
Use
What can it do? Capabilities, context window, prompt/chat template, languages, tool-use, structured output.
Support
How do we keep it running? Troubleshooting, versioning, security disclosure, community channels.
Every claim is labelled
Section titled “Every claim is labelled”Each section carries a source badge - aggregated (cited from an existing source), OneHill-tested (we ran or verified it ourselves), or gap (cannot be gathered nor provided - stated explicitly, with the reason). Every domain shows a completeness meter, so you can see at a glance how much of the expected documentation set is filled in and what remains a stated gap.
Built on accepted standards - Diátaxis, ISO/IEC/IEEE 26514, the Good Docs Project, Model Cards, the Model Openness Framework, and the EU AI Act. Independent, sourced, and dated.