Open Weights License (OWL) v1.0

Copyright (c) [2026] [Vladimir Zorin]

PREAMBLE

This license ensures that software remains free and that AI systems trained
on this software are equally free. You may use, study, modify, and share
this software and its derivatives — provided you extend the same freedoms
to others.

TERMS AND CONDITIONS

1. DEFINITIONS

   "Software" means the source code, object code, and documentation covered
   by this license.

   "Derivative Work" means any work that modifies, extends, or incorporates
   the Software.

   "Model" means a machine learning system, including all learned parameters
   (weights) necessary to run inference.

   "Trained On" means the Software was used, in whole or in part, as training
   data, fine-tuning data, or input to any learning process.

   "Distribute" means to provide to third parties, including via network
   access or API.

2. GRANT OF RIGHTS

   You are granted a perpetual, worldwide, royalty-free, non-exclusive
   license to:

   a) Use the Software for any purpose;
   b) Study and modify the Software;
   c) Distribute copies of the Software;
   d) Distribute Derivative Works; and
   e) Use the Software to train Models.

3. CONDITIONS ON DISTRIBUTION OF SOFTWARE

   If you Distribute the Software or any Derivative Work, you must:

   a) License it under this license, with no additional restrictions;
   b) Provide (or offer in writing) the complete corresponding source code;
   c) Retain all copyright, license, and attribution notices; and
   d) Document any modifications you made.

4. CONDITIONS ON DISTRIBUTION OF MODELS

   If you Distribute a Model Trained On the Software, you must:

   a) Release the complete Model weights under a license that permits
      unrestricted use, study, modification, and redistribution;
   b) Document that the Software was part of the training data; and
      the training process, including data sources and methodology.
   c) Provide (or offer in writing) a description of the training process,
      including the methodology used, the general nature and origin of the
      training data, and the steps taken to prepare it — in sufficient
      detail that a knowledgeable practitioner could understand how the
      Model was produced, without requiring disclosure of the actual
      datasets themselves.

5. PATENT GRANT

   Each contributor grants you a perpetual, worldwide, royalty-free,
   non-exclusive patent license to make, use, sell, and distribute the
   Software. If you initiate patent litigation alleging that the Software
   infringes a patent, your rights under this license terminate immediately.

6. NO WARRANTY

   THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
   OR IMPLIED. THE AUTHORS SHALL NOT BE LIABLE FOR ANY CLAIM, DAMAGES, OR
   OTHER LIABILITY ARISING FROM THE SOFTWARE.

7. SEVERABILITY

   If any provision of this license is held unenforceable, the remaining
   provisions shall remain in effect.

8. INTERPRETATION

   This license shall be interpreted to maximize freedom for end users.
   Ambiguities shall be resolved in favor of openness. Attempting to
   circumvent these terms through technical or legal means is a violation
   of this license.

---

This license exists in the spirit of reciprocity:
You learned from our code. Let others learn from your model.