Open-weights model
A model whose trained weights are published for anyone to download and run. The licence attached to the weights decides what you may do with them, and the training data and code are often not released, so an open-weights model is not necessarily open source.
Why it matters
Open-weights models let organisations run AI in their own environment, which helps with data residency, version control and predictable cost. The gap to the best proprietary models has narrowed: Artificial Analysis put it at 6 points on its Intelligence Index in April 2026, though it remained wider on the hardest reasoning and agentic coding tests.
Licences vary widely. Artificial Analysis groups open-weights models as permissive, commercial use restricted or non-commercial. Some licences add user thresholds, acceptable use policies or conditions on derivatives, so each one needs a review before engineering work starts.
In practice
For example, a UK building society considering a self-hosted model for complaint summaries might shortlist two open-weights models, then rule one out because its licence covers research use only.
Where Rodan fits
Rodan builds AI into controlled, auditable environments in Governed Workflow Platforms work. See also how to check an open-weights model licence before you build on it.