PaddleOCR document parsing now defaults to VL 1.6

What changed

The hosted PaddleOCR preset now uses PP-OCRv6 for image OCR and PaddleOCR-VL-1.6 for document parsing. The image model is unchanged; the document model previously defaulted to PaddleOCR-VL-1.5.

Why this matters to the user

New PaddleOCR document-processing jobs use the newer VL 1.6 model unless the user has configured a custom capability model override. Existing provider tasks continue with the model used when they were submitted.

What the user should do

Nothing is required. Users who need a different PaddleOCR model can keep or set a custom model override.

Notes for release manager

The installed PaddleOCR API SDK supports both defaults and also uses them when callers omit the model.