Quick start
Why WEB Classic
Three deliberate choices:- WEB Classic, not the British or Protestant editions. The Classic edition (
eng-web) renders God’s proper name as “Yahweh” instead of “LORD,” matching the Urantia Papers’ usage in Papers 96–97. - Includes deuterocanon. WEB’s ecumenical edition has Tobit, Judith, Sirach, Wisdom, Baruch (with the Letter of Jeremiah as chapter 6), Greek Daniel (with Prayer of Azariah, Susanna, and Bel and the Dragon embedded in context), and the Maccabees series.
- Public domain. No license restrictions. The only constraint is that the name “World English Bible” is reserved for faithful copies.
Book codes (OSIS)
We use OSIS book codes throughout:Gen, Matt, 1Macc, DanGr, etc.: short, machine-friendly, standardized across CrossWire.
The endpoint accepts OSIS, USFM (GEN), full names (Genesis), and common aliases, all case-insensitive and tolerant of hyphens/underscores:
letterofjeremiah and epjer both return Baruch (since the Letter of Jeremiah is Baruch chapter 6). susanna and belandthedragon both return Greek Daniel.
Endpoints
List all 81 books, with chapter and verse counts
Get a single book
Get a chapter
Get a single verse
Deuterocanon
The deuterocanon flag is exposed ascanon: "deuterocanon" so consumers can filter:
Errors
All errors follow RFC 9457 Problem Details:404for unknown book codes (/bible/NotABook)404for missing chapters (/bible/Gen/999)404for missing verses (/bible/Gen/1/9999)
Source attribution
- Translation: World English Bible (WEB) Classic edition, public domain
- Source: eBible.org, package
eng-web, snapshot date 2026-04-23 - Editor: Michael Paul Johnson and the WEB team
Cross-references and search
Five query surfaces span the cross-corpus space:
Live searches use 1536-d 3-small for HNSW indexability (pgvector caps HNSW at 2000-d for the
vector type). Pre-computed cross-references use 3072-d 3-large for the +8pt empirical quality benefit on Bible-style retrieval. That is fine for batch compute and too slow for live queries.
You can combine includes: ?include=entities,bibleParallels,urantiaParallels returns all three on a single round-trip.
Bible semantic search
Free-form natural-language search across all 17,641 Bible chunks. Each result includes the top-N pre-computed Urantia paragraphs related to that chunk, so a single round-trip surfaces both the Bible matches and the relevant UB content.canon:"ot"|"deuterocanon"|"nt": restrict to a part of the canonbookCode: any OSIS, USFM, full name, or alias: restrict to a single bookurantiaParallelLimit: 0–10, default 3: how many UB paragraphs to attach per Bible chunk (0 suppresses)
UB paragraph → UB paragraphs
Cross-references: UB ↔ Bible parallels
Pre-computed semantic parallels between every Urantia paragraph and the Bible, in both directions. ~146,000 UB→Bible rows + ~176,000 Bible→UB rows, top-10 nearest neighbors per source. Generated with OpenAI’stext-embedding-3-large (3072-d) and exact KNN.
Get the Bible verses related to a UB paragraph
Add?include=bibleParallels to any single-paragraph endpoint:
bibleParallels array with up to 10 entries:
Get the UB paragraphs related to a Bible verse (reverse query)
How to read them
Every Urantia paragraph has its Bible parallels, found by meaning with one method. The method finds passages on the same subject also when they share no words. Each row carriessimilarity, source: "semantic", and embeddingModel, so you can filter by strength and check each result. The model reads common religious words (“Father”, “Spirit”, “Son”) in their usual sense, so a low score can be a match in words only. See How the links are made for the method and for what the number means.
An example of a close match: Matt 5:3 (Beatitudes) → UB 140:3.3 (“Happy are the poor in spirit, the humble”) at similarity 0.854.
