License and compatibility clarification for PP-OCRv5 Japanese OCR model files

[Ryuichi-Ishizawa]

Hello PaddleOCR maintainers,

We are preparing an isolated, offline evaluation of Japanese OCR using fictional, non-sensitive documents. We will not attach documents, source code, credentials, internal addresses, or production data. Before obtaining any files, could you please clarify the following for the exact PaddleOCR 3.7.0 / PP-OCRv5 configuration?

1. For each of PP-OCRv5_mobile_det_infer.tar, PP-OCRv5_mobile_rec_infer.tar, and PP-LCNet_x0_25_textline_ori_infer.tar, who owns the copyright and what license and NOTICE terms apply to the archive and extracted model files?
2. What license and NOTICE terms apply to the Japanese dictionary or character list used by PP-OCRv5_mobile_rec? Please identify the exact filename and version.
3. Is internal commercial use on our own servers permitted? Please distinguish that from redistribution of the model archives, extracted weights, dictionaries, or a service that bundles them.
4. Does the PaddleOCR repository Apache-2.0 license apply to these distributed weights and dictionaries, or are they subject to separate terms?
5. Are the model archives hosted under the official paddle3.0.0 path officially supported with PaddleOCR 3.7.0? Please provide the exact release/commit pairing and an official SHA-256 manifest if available.
6. What local/offline configuration disables automatic downloads and identifies all required files for Japanese OCR?

A link to the relevant official license/NOTICE/model-card document is sufficient. We are requesting information only and are not asking you to review confidential material. Thank you.