Z.aiGLM OCR
GLM-OCR
A compact OCR model with a separate document-layout stage.
Practical use
Practical guidance. Check page order and region boundaries as well as character accuracy. A correct transcription of the wrong table region still produces an incorrect document result. Keep layout coordinates with extracted fields.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API, Downloadable |
| Model identifiers | glm-ocrzai-org/GLM-OCR |
| Size | 0.9 billion parameters, including a GLM-0.5B language decoder.GLM-OCR official documentation |
| Architecture | CogViT image encoder, connector with token downsampling, and layout analysis using PP-DocLayout-V3.GLM-OCR official documentation |
| Licence | MIT.GLM-OCR official documentation |
| API | Current catalogue includes layout parsing.Current Z.ai API catalogue |