Case law
Supreme Court and High Court judgments with parties, citations, benches, and PDF provenance.
Open Legal Data Initiative
The public data layer for Indian legal AI
Case law, statutes, rules, and notifications — structured, licensed, and ready for search, RAG, benchmarks, and production systems. Released on Hugging Face under Apache 2.0.
- Judgment records
- 17.1M
- Legal documents
- 35,851
- Courts
- SC + 25 HCs
- License
- Apache 2.0
Coverage
Statutes & instruments
Acts, Rules, Notifications, Circulars, Orders, Regulations, and guidelines — Central and state.
Legal metadata
Court, jurisdiction, authority, dates, disposition, and quality flags as first-class columns.
Structured corpora
Parquet tables for analytics, streaming loaders, and Dataset Viewer — not a pile of PDFs.
Public datasets
Two live releases on Hugging Face. More annotated corpora ship through Pro and Enterprise.
Case lawApache-2.0Parquet
Indian Case Laws
Structured metadata for Indian court judgments, with provenance to source JSON and original PDFs. Built for search, retrieval, ranking, RAG, and legal NLP evaluation.
- Records
- 17.1 million
- Size
- 10.9 GB
- Timeframe
- 1950–2026
- Coverage
- Supreme Court of India + 25 High Courts
- 17.1M judgment records with parties, docket, CNR, bench, and disposition
- Neutral citations and law-report citations where available
- Parser diagnostics and quality flags for safe filtering
- Links to original eCourts judgment PDFs
case_title, parties, docket_number, cnr_number
neutral_citation, law_report_citation
court_name, bench_name, presiding_judge, coram
decision_date, disposition_text
source_json_s3_url, source_pdf_s3_url
indexable_text, headnote_text, quality_json
Open on Hugging FaceKanoonGPT/indian-case-laws
Statutes & documentsApache-2.0Parquet
Indian Legal Documents
AI-ready Acts, Rules, Notifications, Circulars, Orders, Regulations, and guidelines — cleaned text plus filterable metadata for statutory RAG and legal NLP.
- Records
- 35,851
- Size
- 471 MB
- Timeframe
- 1970–2026
- Coverage
- Central + major states (Maharashtra, UP, Karnataka, and more)
- 35,851 documents with cleaned full text
- Rules, Notifications, and Acts are the largest types
- Central plus state jurisdictions in one table
- Title, type, jurisdiction, authority, and issue date as columns
doc_id, document_title, document_type
document_jurisdiction, issuing_authority
issue_date, short_description
text (cleaned full document)
Open on Hugging FaceKanoonGPT/indian-legal-documents
How access works
Open data for training and research. Annotated inference in Pro. Licensed training data for enterprises.
Research & training
Open data
Free
- Apache-2.0 datasets on Hugging Face
- Case law metadata and statutory documents
- Parquet + Dataset Viewer + streaming loaders
Product inference
Pro
From ₹129
- Annotated data for inference in KanoonGPT Pro
- Section-wise Bare Acts, chat, and KanoonFM
- Same account on web and mobile
Training & licensing
Enterprise
Custom
- Annotated corpora for commercial training
- Licensing and volume terms
- Support for production legal AI systems
Built for teams who ship
Legal search & RAG
Retrieve the right judgment or statute, then cite section and court context.
Benchmarks & eval
Score Indian legal NLP models on real courts, acts, and notifications.
Classification
Type, jurisdiction, bench, and disposition labels for analytics pipelines.
Copilots & agents
Give agents a public, structured Indian-law layer instead of scraping gazettes.
Responsible use
- Verify important legal facts against official gazettes, eCourts PDFs, and court records.
- Do not treat dataset rows or model outputs as legal advice.
- Metadata is machine-extracted and may be incomplete or noisy — filter before high-stakes use.
- Add citations and human review in any product that answers legal questions.
Need annotated data or a commercial license?
Open datasets stay public. Talk to us for enterprise training data, or start Pro for product inference.