AI Model & Third-Party Notices

Conditions: The model card declares Apache-2.0. Preserve the license and required notices when distributing covered materials; modified files must be marked as required by Apache-2.0.

Restrictions: No model-specific additional use restriction was verified in the checked model card. Comply with applicable law and service policy.

Training data: Training-data rights and provenance were not established by this review.

Generated outputs: The model repository license does not establish ownership, copyrightability, or non-infringement of generated outputs.

Redistribution notice: Include a copy of Apache-2.0 and retain applicable copyright, patent, trademark, and attribution notices when redistributing covered materials.

Research note: Model-card license declaration checked. Separate inference-code license was not verified; the official GitHub repository is linked from the card. No separate output or training-data rights conclusion.

Sources checked 2026-10-22; status partially verified.

Conditions: The model card declares Apache-2.0. Preserve the license and required notices when distributing covered materials; modified files must be marked as required by Apache-2.0.

Restrictions: No model-specific additional use restriction was verified in the checked model card. Comply with applicable law and service policy.

Training data: Training-data rights and provenance were not established by this review.

Generated outputs: The model repository license does not establish ownership, copyrightability, or non-infringement of generated outputs.

Redistribution notice: Include a copy of Apache-2.0 and retain applicable copyright, patent, trademark, and attribution notices when redistributing covered materials.

Research note: Card describes this as the full-capacity, undistilled model. Separate code license, training-data rights, and output rights were not established.

Sources checked 2026-10-22; status partially verified.

Conditions: The Schnell model license is Apache-2.0. The Hugging Face repository is gated; comply with its access process and preserve Apache notices when redistributing covered materials.

Restrictions: Repository access is gated and may require acceptance of its conditions. This registry does not establish terms for third-party inference APIs.

Training data: Training-data rights and provenance were not established by this review.

Generated outputs: The model license does not determine rights in generated outputs.

Redistribution notice: Include a copy of Apache-2.0 and retain applicable copyright, patent, trademark, and attribution notices when redistributing covered materials.

Research note: Weights license verified from BFL's official Schnell license and HF record. The repository is gated. Separate code license, training data, and output terms are unknown.

Sources checked 2026-10-22; status partially verified.

Conditions: The official model README declares Apache-2.0. Preserve the license and required notices when distributing covered materials.

Restrictions: No model-specific additional use restriction was verified in the checked model card.

Training data: Training-data rights and provenance were not established by this review.

Generated outputs: The model license does not establish ownership, copyrightability, or non-infringement of generated outputs.

Redistribution notice: Include a copy of Apache-2.0 and retain applicable notices when redistributing covered materials.

Research note: The model README states that Qwen-Image is licensed under Apache 2.0. Training-data rights and output rights remain unverified.

Sources checked 2026-10-22; status partially verified.

Conditions: The model card declares Apache-2.0. Preserve the license and required notices when distributing covered materials.

Restrictions: No additional variant-specific restriction was verified in the checked model card.

Training data: Training-data rights and provenance were not established by this review.

Generated outputs: The model license does not establish output rights.

Redistribution notice: Include a copy of Apache-2.0 and retain applicable notices when redistributing covered materials.

Research note: Exact Diffusers variant in the generator was matched to this repository ID. Separate code, training-data, and output rights were not verified.

Sources checked 2026-10-22; status partially verified.

Conditions: The model card declares Apache-2.0. Preserve the license and required notices when distributing covered materials.

Restrictions: No additional variant-specific restriction was verified in the checked model card.

Training data: Training-data rights and provenance were not established by this review.

Generated outputs: The model license does not establish output rights.

Redistribution notice: Include a copy of Apache-2.0 and retain applicable notices when redistributing covered materials.

Research note: Exact Diffusers variant in the generator was matched to this repository ID. Separate code, training-data, and output rights were not verified.

Sources checked 2026-10-22; status partially verified.

Conditions: The publisher states code and weights are Apache-2.0. Model-card guidance calls for verification mechanisms, disclosure of AI involvement, and permissions for protected styles; treat these as publisher guidance, not additional Apache clauses.

Restrictions: Publisher guidance: implement verification mechanisms, disclose AI involvement, and obtain permission when adapting protected styles.

Training data: Training-data rights and provenance were not independently verified.

Generated outputs: Commercial inference license does not settle ownership or clearance for generated songs or lyrics.

Redistribution notice: Include a copy of Apache-2.0 and preserve the copyright notice and disclaimer on redistribution.

Output disclosure: Disclose that AI was involved in creating the output. The publisher does not prescribe exact wording; this is model-card responsible-use guidance, not an Apache-2.0 notice condition.

Research note: README expressly covers both code and weights under Apache-2.0. The normal plan-based routing inspected does not select this worker.

Sources checked 2026-10-22; status verified.

Conditions: README and LICENSE say Apache-2.0, but Hugging Face model metadata says stable-audio-community. Do not rely on commercial deployment or redistribution until the publisher resolves this discrepancy in writing.

Restrictions: Do not approve commercial use or weight redistribution while repository license metadata conflicts.

Training data: Training-data rights and provenance were not independently verified.

Generated outputs: Output rights cannot be concluded from the conflicting repository metadata.

Output disclosure: The model README recommends disclosure of AI involvement; no exact wording was verified. The license conflict remains unresolved.

Research note: Material conflict between repository README/LICENSE (Apache-2.0) and Hugging Face API metadata (stable-audio-community). Obtain written clarification; do not infer.

Sources checked 2026-10-22; status conflicting.

Conditions: Publisher explicitly states that the Apache-licensed weights may be deployed in production and commercial APIs. Preserve Apache notices when distributing covered materials.

Restrictions: Resolve the runtime checkpoint version because the wrapper does not explicitly pass the repository ID. No separate output terms verified.

Training data: Model card identifies CC BY 3.0 and CC BY 4.0 audio in v1.0 training data; this review did not establish downstream obligations or rights for the complete dataset.

Generated outputs: The model card's commercial deployment statement is not a grant of output ownership or a non-infringement warranty.

Redistribution notice: Include a copy of Apache-2.0 and retain applicable notices when redistributing covered weights.

Research note: Weights license and commercial API statement verified; the generator wrapper invokes KPipeline without an explicit model ID, so exact runtime resolution needs engineering verification.

Sources checked 2026-10-22; status partially verified.

Conditions: The model card declares MIT and the linked official code repository is MIT. Preserve the copyright and permission notice when distributing covered materials.

Restrictions: No model-specific commercial inference restriction was verified.

Training data: Training-data rights and provenance were not established.

Generated outputs: The MIT license does not grant output ownership or resolve rights in generated sounds.

Redistribution notice: Include the MIT copyright and permission notice in copies or substantial portions of covered materials.

Research note: The app loads s3_xl and clones the linked source project. Exact weight-bundle notice scope and dependency licenses need review.

Sources checked 2026-10-22; status partially verified.

Conditions: The canonical model card and linked official code repository declare MIT. Preserve the copyright and permission notice when distributing covered materials.

Restrictions: The exact upstream organization spelling is lowercase kandinskylab; use the canonical repository ID above.

Training data: Training-data rights and provenance were not established.

Generated outputs: The model license does not establish output rights.

Redistribution notice: Include the MIT copyright and permission notice in copies or substantial portions of covered materials.

Research note: Canonical HF organization and repository casing resolved. The same upstream code license is linked by the model card.

Sources checked 2026-10-22; status verified.

Conditions: The model card and linked official code repository declare MIT. Preserve the copyright and permission notice when distributing covered materials.

Restrictions: This is a distilled VSR ancillary checkpoint; the exact canonical ID uses lowercase kandinskylab.

Training data: Training-data rights and provenance were not established.

Generated outputs: The model license does not establish output rights.

Redistribution notice: Include the MIT copyright and permission notice in copies or substantial portions of covered materials.

Research note: Canonical ID resolved. Model card describes the distilled VSR variant; redistribution scope for all bundled files still requires component review.

Sources checked 2026-10-22; status verified.

Conditions: The official model card and code repository declare MIT. The model README states: “You can strictly use the generated music for commercial purposes.” This is the publisher's statement, not a guarantee of ownership or non-infringement.

Restrictions: The 1.7B LM is a subdirectory, not a standalone repository in the checked sources. Do not assume a subcomponent has separate terms.

Training data: Training-data rights and provenance were not independently established.

Generated outputs: Publisher states generated music may be used commercially; ownership, third-party rights, originality, and non-infringement are not established by that statement.

Redistribution notice: Include the MIT copyright and permission notice in copies or substantial portions of covered materials.

Research note: Main checkpoint resolution and 1.7B LM subfolder are inferred from the checked application config and official parent model zoo; confirm the exact downloaded files and license scope. 4B LM is documented as an alternative but not configured in the inspected active path.

Sources checked 2026-10-22; status partially verified.

Conditions: The model card declares Apache-2.0, but a standalone LICENSE file was not located in the reviewed repository. Obtain publisher confirmation and inspect downloaded files before production approval or redistribution.

Restrictions: A citation section names Shreyas Dixit but was not stated as an additional license condition.

Training data: Training-data provenance and rights were not established.

Generated outputs: Output rights and commercial-use conditions were not independently established.

Research note: HF model-card metadata/front matter declares Apache-2.0. No separate LICENSE file was verified; publisher clarification is needed before legal approval.

Sources checked 2026-10-22; status partially verified.

Conditions: The model card declares Apache-2.0, but a standalone LICENSE file was not located in the reviewed repository. Obtain publisher confirmation and inspect downloaded files before production approval or redistribution.

Restrictions: The README describes training on 250 YouTube videos; their licensing and reuse rights were not established.

Training data: Training-data sources are described as 250 YouTube videos, but licensing, permissions, and rights were not verified.

Generated outputs: Output rights and commercial-use conditions were not independently established.

Research note: HF model-card metadata declares Apache-2.0, but no standalone LICENSE file was verified. Publisher clarification is needed; training-data rights are unresolved.

Sources checked 2026-10-22; status partially verified.

Conditions: The model card declares Apache-2.0, but a standalone LICENSE file was not located in the reviewed repository. Obtain publisher confirmation and inspect downloaded files before production approval or redistribution.

Restrictions: The README describes a dataset of 5,000 YouTube videos; their licensing and reuse rights were not established.

Training data: Training-data sources are described as 5,000 YouTube videos, but licensing, permissions, and rights were not verified.

Generated outputs: Output rights and commercial-use conditions were not independently established.

Research note: HF model-card metadata declares Apache-2.0, but no standalone LICENSE file was verified. Publisher clarification is needed; training-data rights are unresolved.

Sources checked 2026-10-22; status partially verified.