LimiX-2 / LICENSE

Update LICENSE

bf1b0c9 verified 2 months ago

10.3 kB

		StableAI LimiX Non-Commercial License Version 1.0
						September 2026

	Copyright © 2026 Stable AI Technology Co., Ltd. 
	Copyright © 2026 Wen Zhun (Xiong’an) Technology Co., Ltd. 
	Copyright © 2026 稳准智能(雄安)科技有限公司

	The foregoing names identify the same copyright owner and are
	alternative Chinese and English names of the same legal entity.

	IMPORTANT NOTICE: This License is a custom non-commercial model license.
	It is not an open source license and does not grant commercial usage
	rights.


	1.  Definitions
		“Licensor” means Stable AI Technology Co., Ltd., including its
		applicable Chinese and English corporate names identified above.

		“Model” means the StableAI LimiX model and associated materials
		distributed by the Licensor under this License, including but not
		limited to model weights, parameters, checkpoints, tokenizer files,
		architecture implementation, inference code, fine-tuning code,
		evaluation code, configuration files, documentation, embeddings,
		representations, and other materials designated by the Licensor.

		“Derivative Model” means any model, adapter, module, checkpoint, or
		system created, developed, trained, fine-tuned, distilled, adapted,
		modified, or otherwise derived from the Model or a substantial portion
		of the Model.

		Derivative Models include but are not limited to: (a) fine-tuned
		checkpoints; (b) LoRA adapters and parameter-efficient fine-tuning
		modules; (c) prompt-tuned models; (d) retrained models; (e) distilled
		models; (f) models incorporating Model parameters, representations,
		embeddings, or learned capabilities.

		“Outputs” means predictions, generated results, analyses, embeddings,
		classifications, representations, intermediate results, or other
		information generated by the Model.

		“Distribution” means providing access to the Model or any Derivative
		Model to a third party.

		“Non-Commercial Purpose” means use solely for academic research,
		scientific research, education, evaluation, benchmarking, personal
		experimentation, non-commercial data science competitions, and
		non-commercial demonstrations.

		Research conducted by an entity primarily engaged in commercial
		activities shall not qualify as Non-Commercial Purpose if such research
		is intended to support commercial products, services, operations,
		strategic development, or other commercial objectives.

		“Commercial Purpose” means any use intended to obtain commercial
		advantage, directly or indirectly, monetary compensation, business
		benefit, or operational advantage, including production deployment,
		commercial products, customer-facing applications, consulting services,
		internal business operations, SaaS, API services, hosted inference,
		managed services, and commercial agentic systems.

		“Foundation Model” means a large-scale machine learning model designed
		to perform general-purpose or domain-specific learning through
		large-scale pretraining, in-context learning, representation learning,
		or similar paradigms.

		“Competing Foundation Model” means a Foundation Model whose primary
		purpose includes structured-data reasoning, prediction, representation
		learning, or in-context learning over: (a) tabular data; (b) structured
		data; (c) time-series data; (d) graph data; (e) other structured or
		semi-structured data domains.

		Competing Foundation Models include large-scale foundation model
		architectures or systems based on: (a) tabular foundation models; (b)
		structured-data foundation models; (c) time-series foundation models;
		(d) graph foundation models designed for large-scale pretraining or
		in-context learning; (e) in-context learning or similar paradigms; (f)
		mixture-of-experts architectures providing substantially similar
		capabilities; (g) agentic systems or model orchestration systems
		providing substantially similar structured-data reasoning or prediction
		capabilities.


	2.  License Grant
		Subject to this License, the Licensor grants You a worldwide,
		non-exclusive, royalty-free license to use the Model solely for
		Non-Commercial Purposes.

		You may: (a) use the Model for research, education, evaluation, and
		benchmarking; (b) create Derivative Models; (c) distribute Derivative
		Models subject to this License.

		No commercial rights are granted.


	3.  Distribution of Model and Derivative Models
		You may distribute the Model or Derivative Models solely for
		Non-Commercial Purposes.

		Any distribution must: (a) include this License or a publicly accessible
		link to this License; (b) include attribution to StableAI LimiX; (c)
		disclose material modifications; (d) state that the distributed model is
		derived from StableAI LimiX.

		Any Derivative Model distributed under this License shall remain subject
		to the same restrictions, limitations, and obligations as the Model.

		You may not distribute the Model or any Derivative Model under terms
		that remove or weaken the restrictions of this License.


	4.  Fine-Tuned Models and Checkpoints
		Fine-tuned checkpoints, LoRA adapters, and other Derivative Models may
		be distributed only under this License.

		Recipients of Derivative Models do not receive any rights for Commercial
		Purposes.

		You may not sublicense, relicense, or otherwise distribute the Model or
		Derivative Models under terms that: (a) permit Commercial Purpose; (b)
		remove Non-Commercial restrictions; (c) eliminate attribution
		obligations; (d) prevent enforcement of this License.


	5.  Competition and Benchmark Usage
		You may use the Model in non-commercial competitions, including Kaggle
		competitions, provided that: (a) the use complies with competition
		rules; (b) no commercial service is created; (c) no paid access is
		provided; (d) no official endorsement by StableAI is implied.


	6.  Personal Demonstrations
		Personal, educational, and research demonstrations are permitted if
		they: (a) do not generate revenue; (b) do not provide commercial
		services; (c) do not expose Model capabilities through SaaS, API, hosted
		inference, remote access services, or similar services.


	7.  Commercial Restrictions
		You may not use the Model, Outputs, Derivative Models, embeddings,
		representations, or information derived from the Model for any
		Commercial Purpose.

		Commercial use requires a separate commercial agreement with the
		Licensor.


	8.  SaaS, API, MoE, and Agentic Service Restrictions
		You may not use the Model or Derivative Models to provide: (a) API
		access; (b) hosted inference; (c) model-as-a-service; (d) managed
		services; (e) remote access services; (f) commercial mixture-of-experts
		systems; (g) commercial agentic AI systems; (h) commercial model
		orchestration systems; (i) any system that provides third-party access
		to Model capabilities for Commercial Purposes.

		Non-commercial academic experimentation involving such architectures is
		permitted.


	9.  Outputs and Competitive Restrictions
		The Licensor does not claim ownership of Outputs.

		You may not use Outputs, embeddings, representations, intermediate
		states, evaluations, or information derived from the Model to train,
		fine-tune, distill, evaluate, optimize, or improve any Competing
		Foundation Model for commercial or competitive purposes.

		You may not use the Model or Outputs to: (a) select, curate, filter,
		label, or evaluate training data for a Competing Foundation Model; (b)
		generate, evaluate, rank, or improve synthetic training data for a
		Competing Foundation Model; (c) construct evaluation pipelines for a
		Competing Foundation Model; (d) design, optimize, or improve training
		pipelines for a Competing Foundation Model.


	10. Competing Foundation Model Restrictions
		You may not use the Model, Outputs, representations, evaluations, or
		information derived from the Model to develop, improve, optimize, or
		support a Competing Foundation Model.

		Non-commercial academic research involving analysis of model
		capabilities is permitted.

		Commercial development, optimization, deployment, or integration of
		competing models is prohibited.


	11. Attribution Requirements
		If You distribute the Model, a Derivative Model, publicly demonstrate
		the Model, or publicly disclose research based on the Model, You must
		provide attribution:

		“Built with StableAI LimiX”


	12. Trademarks
		This License does not grant permission to use StableAI, LimiX, or other
		Licensor trademarks except for required attribution.

		You may not imply endorsement, sponsorship, partnership, certification,
		or official affiliation.


	13. No Commercial Relicensing
		The Model and Derivative Models may not be sublicensed, relicensed, or
		distributed under any terms that: (a) remove commercial restrictions;
		(b) eliminate attribution requirements; (c) weaken enforcement of this
		License.


	14. Acceptable Use Policy
		You may not use the Model for: (a) unlawful activities; (b) privacy
		violations; (c) unauthorized surveillance; (d) fraud or deceptive
		activities; (e) harmful or abusive applications; (f) activities
		violating applicable laws; (g) security exploitation or unauthorized
		access; (h) high-impact decisions affecting individuals without
		appropriate human oversight.


	15. Disclaimer of Warranty
		THE MODEL IS PROVIDED “AS IS” WITHOUT WARRANTIES OR CONDITIONS OF ANY
		KIND.


	16. Limitation of Liability
		To the maximum extent permitted by law, the Licensor shall not be liable
		for damages arising from use of the Model.


	17. Termination
		Your rights automatically terminate upon violation of this License.

		Upon termination, You must cease all use and Distribution of the Model
		and Derivative Models.

		Rights may be reinstated if the violation is cured within thirty (30)
		days after written notice from the Licensor.


	18. Governing Law and Severability
		This License shall be governed by and construed in accordance with the
		laws selected by the Licensor and specified in the applicable licensing
		documentation.

		If any provision of this License is held invalid or unenforceable, the
		remaining provisions shall remain effective.


	19. Commercial Licensing
		Commercial use requires a separate commercial license agreement with
		StableAI.


	END OF LICENSE

	StableAI LimiX Non-Commercial License Version 1.0
	September 2026

	Copyright © 2026 Stable AI Technology Co., Ltd.
	Copyright © 2026 Wen Zhun (Xiong’an) Technology Co., Ltd.
	Copyright © 2026 稳准智能(雄安)科技有限公司

	The foregoing names identify the same copyright owner and are
	alternative Chinese and English names of the same legal entity.

	IMPORTANT NOTICE: This License is a custom non-commercial model license.
	It is not an open source license and does not grant commercial usage
	rights.


	1. Definitions
	“Licensor” means Stable AI Technology Co., Ltd., including its
	applicable Chinese and English corporate names identified above.

	“Model” means the StableAI LimiX model and associated materials
	distributed by the Licensor under this License, including but not
	limited to model weights, parameters, checkpoints, tokenizer files,
	architecture implementation, inference code, fine-tuning code,
	evaluation code, configuration files, documentation, embeddings,
	representations, and other materials designated by the Licensor.

	“Derivative Model” means any model, adapter, module, checkpoint, or
	system created, developed, trained, fine-tuned, distilled, adapted,
	modified, or otherwise derived from the Model or a substantial portion
	of the Model.

	Derivative Models include but are not limited to: (a) fine-tuned
	checkpoints; (b) LoRA adapters and parameter-efficient fine-tuning
	modules; (c) prompt-tuned models; (d) retrained models; (e) distilled
	models; (f) models incorporating Model parameters, representations,
	embeddings, or learned capabilities.

	“Outputs” means predictions, generated results, analyses, embeddings,
	classifications, representations, intermediate results, or other
	information generated by the Model.

	“Distribution” means providing access to the Model or any Derivative
	Model to a third party.

	“Non-Commercial Purpose” means use solely for academic research,
	scientific research, education, evaluation, benchmarking, personal
	experimentation, non-commercial data science competitions, and
	non-commercial demonstrations.

	Research conducted by an entity primarily engaged in commercial
	activities shall not qualify as Non-Commercial Purpose if such research
	is intended to support commercial products, services, operations,
	strategic development, or other commercial objectives.

	“Commercial Purpose” means any use intended to obtain commercial
	advantage, directly or indirectly, monetary compensation, business
	benefit, or operational advantage, including production deployment,
	commercial products, customer-facing applications, consulting services,
	internal business operations, SaaS, API services, hosted inference,
	managed services, and commercial agentic systems.

	“Foundation Model” means a large-scale machine learning model designed
	to perform general-purpose or domain-specific learning through
	large-scale pretraining, in-context learning, representation learning,
	or similar paradigms.

	“Competing Foundation Model” means a Foundation Model whose primary
	purpose includes structured-data reasoning, prediction, representation
	learning, or in-context learning over: (a) tabular data; (b) structured
	data; (c) time-series data; (d) graph data; (e) other structured or
	semi-structured data domains.

	Competing Foundation Models include large-scale foundation model
	architectures or systems based on: (a) tabular foundation models; (b)
	structured-data foundation models; (c) time-series foundation models;
	(d) graph foundation models designed for large-scale pretraining or
	in-context learning; (e) in-context learning or similar paradigms; (f)
	mixture-of-experts architectures providing substantially similar
	capabilities; (g) agentic systems or model orchestration systems
	providing substantially similar structured-data reasoning or prediction
	capabilities.


	2. License Grant
	Subject to this License, the Licensor grants You a worldwide,
	non-exclusive, royalty-free license to use the Model solely for
	Non-Commercial Purposes.

	You may: (a) use the Model for research, education, evaluation, and
	benchmarking; (b) create Derivative Models; (c) distribute Derivative
	Models subject to this License.

	No commercial rights are granted.


	3. Distribution of Model and Derivative Models
	You may distribute the Model or Derivative Models solely for
	Non-Commercial Purposes.

	Any distribution must: (a) include this License or a publicly accessible
	link to this License; (b) include attribution to StableAI LimiX; (c)
	disclose material modifications; (d) state that the distributed model is
	derived from StableAI LimiX.

	Any Derivative Model distributed under this License shall remain subject
	to the same restrictions, limitations, and obligations as the Model.

	You may not distribute the Model or any Derivative Model under terms
	that remove or weaken the restrictions of this License.


	4. Fine-Tuned Models and Checkpoints
	Fine-tuned checkpoints, LoRA adapters, and other Derivative Models may
	be distributed only under this License.

	Recipients of Derivative Models do not receive any rights for Commercial
	Purposes.

	You may not sublicense, relicense, or otherwise distribute the Model or
	Derivative Models under terms that: (a) permit Commercial Purpose; (b)
	remove Non-Commercial restrictions; (c) eliminate attribution
	obligations; (d) prevent enforcement of this License.


	5. Competition and Benchmark Usage
	You may use the Model in non-commercial competitions, including Kaggle
	competitions, provided that: (a) the use complies with competition
	rules; (b) no commercial service is created; (c) no paid access is
	provided; (d) no official endorsement by StableAI is implied.


	6. Personal Demonstrations
	Personal, educational, and research demonstrations are permitted if
	they: (a) do not generate revenue; (b) do not provide commercial
	services; (c) do not expose Model capabilities through SaaS, API, hosted
	inference, remote access services, or similar services.


	7. Commercial Restrictions
	You may not use the Model, Outputs, Derivative Models, embeddings,
	representations, or information derived from the Model for any
	Commercial Purpose.

	Commercial use requires a separate commercial agreement with the
	Licensor.


	8. SaaS, API, MoE, and Agentic Service Restrictions
	You may not use the Model or Derivative Models to provide: (a) API
	access; (b) hosted inference; (c) model-as-a-service; (d) managed
	services; (e) remote access services; (f) commercial mixture-of-experts
	systems; (g) commercial agentic AI systems; (h) commercial model
	orchestration systems; (i) any system that provides third-party access
	to Model capabilities for Commercial Purposes.

	Non-commercial academic experimentation involving such architectures is
	permitted.


	9. Outputs and Competitive Restrictions
	The Licensor does not claim ownership of Outputs.

	You may not use Outputs, embeddings, representations, intermediate
	states, evaluations, or information derived from the Model to train,
	fine-tune, distill, evaluate, optimize, or improve any Competing
	Foundation Model for commercial or competitive purposes.

	You may not use the Model or Outputs to: (a) select, curate, filter,
	label, or evaluate training data for a Competing Foundation Model; (b)
	generate, evaluate, rank, or improve synthetic training data for a
	Competing Foundation Model; (c) construct evaluation pipelines for a
	Competing Foundation Model; (d) design, optimize, or improve training
	pipelines for a Competing Foundation Model.


	10. Competing Foundation Model Restrictions
	You may not use the Model, Outputs, representations, evaluations, or
	information derived from the Model to develop, improve, optimize, or
	support a Competing Foundation Model.

	Non-commercial academic research involving analysis of model
	capabilities is permitted.

	Commercial development, optimization, deployment, or integration of
	competing models is prohibited.


	11. Attribution Requirements
	If You distribute the Model, a Derivative Model, publicly demonstrate
	the Model, or publicly disclose research based on the Model, You must
	provide attribution:

	“Built with StableAI LimiX”


	12. Trademarks
	This License does not grant permission to use StableAI, LimiX, or other
	Licensor trademarks except for required attribution.

	You may not imply endorsement, sponsorship, partnership, certification,
	or official affiliation.


	13. No Commercial Relicensing
	The Model and Derivative Models may not be sublicensed, relicensed, or
	distributed under any terms that: (a) remove commercial restrictions;
	(b) eliminate attribution requirements; (c) weaken enforcement of this
	License.


	14. Acceptable Use Policy
	You may not use the Model for: (a) unlawful activities; (b) privacy
	violations; (c) unauthorized surveillance; (d) fraud or deceptive
	activities; (e) harmful or abusive applications; (f) activities
	violating applicable laws; (g) security exploitation or unauthorized
	access; (h) high-impact decisions affecting individuals without
	appropriate human oversight.


	15. Disclaimer of Warranty
	THE MODEL IS PROVIDED “AS IS” WITHOUT WARRANTIES OR CONDITIONS OF ANY
	KIND.


	16. Limitation of Liability
	To the maximum extent permitted by law, the Licensor shall not be liable
	for damages arising from use of the Model.


	17. Termination
	Your rights automatically terminate upon violation of this License.

	Upon termination, You must cease all use and Distribution of the Model
	and Derivative Models.

	Rights may be reinstated if the violation is cured within thirty (30)
	days after written notice from the Licensor.


	18. Governing Law and Severability
	This License shall be governed by and construed in accordance with the
	laws selected by the Licensor and specified in the applicable licensing
	documentation.

	If any provision of this License is held invalid or unenforceable, the
	remaining provisions shall remain effective.


	19. Commercial Licensing
	Commercial use requires a separate commercial license agreement with
	StableAI.


	END OF LICENSE