pypi / rapidocr

rapidocr

What changed in rapidocr 3, from its official release notes, upgrade guides and advisories. Sources checked 2026-10-24.

2 breaking1 security8 new APIs and options1 performance2 behavior and defaults6 fixes and other

See which of these apply to your code, ranked, in your agent:

npx @unstale/cli init

rapidocr 320 changes

OpenCV is headless by default, with GUI support available optionally.
3.10.0release notes
PyTorch 1.13+ is required for secure model loading.
3.10.0release notes

Torch weights load with weights_only=True to prevent code execution.
3.10.0release notes

Images can be decoded from memory addresses.
3.10.0release notes
Adds multilingual model routing and licensing metadata.
3.10.0release notes
Image loading can use raw memory buffers with format and stride metadata.
3.10.0release notes
Adds the use_apply_vertical_padding parameter for controlling vertical padding.
3.9.2release notes
Adds the use_preprocess_img parameter for controlling image preprocessing.
3.9.2release notes
Adds MNN backend support for PP-OCrv6.
3.9.1release notes
Adds PaddlePaddle backend support for PP-OCRv6, allowing developers to use this inference backend.
3.9.1release notes
Adds PyTorch as an inference engine for PP-OCRv6.
3.9.1release notes

Lazy model loading is locked so first use constructs the model once.
3.10.0release notes
Routes are centralized in YAML, and supported languages are exposed.
3.10.0release notes

The function now uses the device it is given.
3.10.0release notes
The detector now honors limit_side_len when limit_type is max.
3.10.0release notes
MNN output extraction uses getNumpyData() instead of getData().
3.10.0release notes
The torch engine now honors model_root_dir for recognition dictionary and string paths.
3.10.0release notes
Arabic OCR initialization validates python-bidi before proceeding.
3.10.0release notes
Adjusts the PP-OCRv6 Paddle model key names in the YAML config.
3.9.1release notes