- Who it helps
- Developers preparing document inputs on machines without a dedicated GPU.
- Availability
- The public implementation supports Python 3.10 and later. Code uses Apache-2.0; model weights have separate terms with eligibility limits and restrictions on competing services. Check those terms for your use.
Useful output
The project documents Markdown text plus structured layout information such as boxes and confidence. Keeping that layout alongside an answer can help a reader return to the source region. OCR errors still need checking.
Read the size claim carefully
The smallest recognizer is only one component. The default system also includes detection and other models, so a headline parameter count does not describe the complete download or runtime.
What you can do
- Inspect the model card and compare extracted text with a small, non-sensitive document before using it in retrieval. Open reference ↗
Where the evidence stops
Documentation was checked without extracting a document. Handwriting, unusual layouts and licensing eligibility remain use-specific constraints.
Sources read. Sources read October 20, 2026. No installation, model inference or complete external workflow was executed.
Sources & further reading
Hugging Face · implementation
Indra OCR model card ↗
Model card documents compact CPU OCR components and separate weight terms. OCR is adjacent infrastructure for document-grounded generation; accuracy was not tested.
Datalab · page checked 2026-10-20
GitHub · implementation
Indra implementation ↗
README documents PDF/image input, text and layout output, CPU execution and separate code/weight licensing.
Datalab · page checked 2026-10-20
Coverage note: Primary evidence establishes the documented scope. Independent execution and complete YouTube, X and community coverage remain unavailable.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Added to Current
Indra extracts text and layout from PDFs and images on a CPU, producing material that a document-grounded AI workflow can inspect.