v1 — Stable Release
February 25, 2026 The Jmail Data API is now stable and ready for production use. Schemas will not change without a version bump.What’s in v1
- 1.78M emails — Full Epstein email archive (Gmail, Yahoo, DOJ Datasets 8-11)
- 1.41M documents — DOJ volumes 8-11, House Oversight, court records, all with metadata
- 531K+ document full-text shards — Extracted text from all DOJ volumes via Reducto
- 18K photos with AI-generated descriptions
- 473 identified people via AWS Rekognition facial recognition
- 4.5K iMessages recovered from device forensic extractions
- 414K crowd-sourced star ratings from jmail.world users
Formats
Every dataset is available in both Parquet (columnar analytics) and NDJSON (gzipped, for streaming). Python client, DuckDB, and direct download all supported.Infrastructure
- Served from Cloudflare R2 — zero egress fees, no rate limits, no API keys
- ETag-based caching for efficient polling
- Content negotiation (extensionless paths redirect to Parquet by default)
/latest/*version alias always points to current version