00 · IN THREE MINUTES

The answer in three steps

  1. 1A model is a bundle of artifacts and rights, not a single file.
  2. 2Open weights may allow local use while withholding training code, data or modification rights.
  3. 3A meaningful claim should state exactly what is released and under which licence.

01 · THE STACK HAS SEVERAL LAYERS

The stack has several layers

Source code runs training and inference; weights contain learned parameters; configuration describes architecture; data and documentation explain how the system was produced. Each layer can be open, restricted or absent.

02 · ACCESS IS NOT PERMISSION

Access is not permission

A public download may come with limits on commercial use, fields of use or redistribution. Open-source definitions focus on freedoms to use, study, modify and share, not merely on whether a file can be obtained.

FIG. 02How the system changes state
A conceptual mechanism map. Geometry, scale and timing are explanatory unless labelled otherwise.

03 · TRAINING DATA IS DIFFICULT BUT IMPORTANT

Training data is difficult but important

Copyright, privacy and scale can prevent full republication of a dataset. Without sufficient information about data and processing, outsiders may be unable to reproduce or fully audit model behavior.

04 · OPENNESS HAS DEGREES

Openness has degrees

Frameworks score which components and documentation are available. This makes comparisons more informative than a binary badge, especially when weights are open but the recipe is not.

05 · OPEN DOES NOT MEAN SAFE OR UNBIASED

Open does not mean safe or unbiased

Transparency can support inspection and adaptation, while the same access can enable misuse. Safety properties, licence freedoms and reproducibility are related questions, not interchangeable ones.

06 · SOURCES AND EVIDENCE

Sources and evidence

Claims are linked to foundational papers, standards or the primary study behind the update.

  1. 01
    The Open Source AI Definition 1.0

    Supports a defined mechanism, measurement or evidence boundary in this article.

    TECHNICAL STANDARD ↗
  2. 02
    The Model Openness Framework

    Supports a defined mechanism, measurement or evidence boundary in this article.

    TECHNICAL STANDARD ↗
CHANGE LOG29 Aug 2026 · First five-language edition; mechanisms, limits, diagrams and sources checked.