Blog
/
Trinity is moving to OpenMDW-1.1

Trinity is moving to OpenMDW-1.1

Aligning Trinity with a licensing standard designed for open AI model distributions.

Today we are moving the entire Trinity family to OpenMDW-1.1, the Linux Foundation's permissive license built for open AI model distributions.

That includes Trinity Nano and Mini, Trinity-Large, Trinity-Large-Thinking, and every quantized variant we release alongside them.

Why OpenMDW

A model release often covers many different aspects of the model pipeline. It can include architecture, weights, configuration, documentation, code, evaluation materials, and sometimes data and training code. Traditional software licenses were written primarily for source code, so applying them across model artifacts can leave ambiguity at the edges.

Apache 2.0 cleanly covers code, but it was never written to speak to weights, configs, evaluation materials, or data as a single distribution. Developers are forced to reason about each artifact separately. OpenMDW fixes that by covering all model materials a release includes under one instrument.

OpenMDW gives developers a clearer answer across the artifacts they use together, while preserving the open posture Trinity has had from the start.

What changes

Every Trinity release now ships under OpenMDW-1.1, and we've updated previously released models to this license.

That means one license across the Trinity family, including Nano, Mini, Large, Large-Thinking, and their quantized variants. It also means simpler compliance for teams that want to build on Trinity, adapt it, redistribute it, or use it in production.

What does not change

Trinity remains open for developers and enterprises to build with.

You can inspect the models, post-train them, host them yourself, distill them, modify them, and deploy them commercially. There are no field-of-use restrictions and no separate commercial license requirement.

Trinity has always been about delivering excellent open models built in the US, and giving developers full ownership of what they build with them. The move to OpenMDW is a cleaner expression of that same commitment.

A shared standard for open models

NVIDIA has announced plans to adopt OpenMDW-1.1 across future releases of its Cosmos, Isaac GR00T, Ising, and Nemotron open model families.

The promise of open models has always been freedom. A shared license makes that freedom easier to exercise. The terms are the same, recognizable, and permissive across the board, so the decision comes down to which model is best for the job rather than which license you are willing to accept. It lowers the cost of choosing open, and it makes the practical advantages of open models, adaptability, ownership, and choice, available without the licensing friction that has historically come with them.

Trinity, as always, is yours to build with.


Explore the Trinity collections

Resources