Download the dataset
The whole corpus, per version, with a sha256 and byte size for every file. Versions are immutable: a citation names a version, and a corrected corpus ships as a new version — the old one keeps resolving. Take only the parts whose licence you can use.
Read the identifier policy for how these URLs stay stable, and the method page for exactly what is text, computed and asserted.
Version 0.8.0current
built 2026-07-25 · 77,881 tokens · 6,236 versesmanifest.json sha256
00ca019ba8cf994001832177dabf43043494d8ba4c314f886879456891ed7a94The complete artifact set for this version in a single archive: text, morphology, translations, numbering, mappings and the manifest. Because it embeds the GPL Leeds morphology (tokens.jsonl and morphology/), the tarball as a whole is GPL-2.0-or-later. Verify it against the published sha256 before use. For CC-BY-only text, take the per-file text group instead.
- quranbench-corpus-v0.8.0.tar.gz
b2a9aad92895e7e8… · 13.7 MB
Uthmani text from the Tanzil Project, plus verse rows, surah metadata, numbering schemes, identifier mappings and the build manifest. Take this if you only want the Quran text under CC-BY. Full verse text lives in verses.jsonl.
- identifiers.json
ad691b421f7bcf70… · 1.8 KB - lexicon/coverage-report.md
e1bffebf0660cbea… · 2.0 KB - lexicon/LANE-ATTRIBUTION.md
cd462cd41ebc8114… · 1.9 KB - lexicon/lane.json
2ac8de704c515a1d… · 18.2 MB - lexicon/LICENSE.md
ad6f0640b00edab1… · 907 B - mapping/mapping.schema.json
26eeb699088f17ad… · 1.7 KB - mapping/v0.7.0-to-v0.8.0.json
ae66e1d08d47c127… · 540 B - numbering/kufan.json
43f70350303e536d… · 906 B - numbering/numbering.schema.json
31dfd3ef638e02b7… · 1.4 KB - sources.json
cd706e7bd35b907a… · 7.8 KB - surahs.json
b4f4992a245c4780… · 93.1 KB - verses.jsonl
987aa84729fc0890… · 5.4 MB
Root, lemma, part-of-speech and segment annotation from the Leeds Quranic Arabic Corpus. tokens.jsonl combines the Quran text with this annotation, so the combined file is GPL. If you cannot accept copyleft, use verses.jsonl from the text group instead.
- morphology/alignment-report.md
e99dee1e032fb2e1… · 2.6 KB - morphology/ATTRIBUTION.md
6582fc699af560b2… · 1.5 KB - morphology/GLOSS-ATTRIBUTION.md
95069e0f497350ad… · 2.5 KB - morphology/gloss-report.md
89869017c8c9c792… · 10.8 KB - morphology/LICENSE
edaef632cbb643e4… · 17.6 KB - morphology/roots.json
b482fea1fa5dafa6… · 2.2 MB - tokens.jsonl
7f014cdf703a1032… · 72.0 MB
Verse-level human translation editions in the public domain — the ones you may redistribute. Correspondence to the Arabic is verse-level only, never word-level. Each edition ships its own LICENSE file. Display-only editions (shown on the site under a licence that permits display but not redistribution, e.g. Talal Itani under CC BY-NC-ND) are deliberately excluded from these downloads and from the full tarball; they are listed below with the reason.
- translations/en-palmer.jsonl
7131a1c14f4ab442… · 1.3 MB - translations/en-palmer.LICENSE.md
82b68ac668616d5c… · 1.1 KB - translations/en-pickthall.jsonl
f84f0a5c532ae0e7… · 1.4 MB - translations/en-pickthall.LICENSE.md
3578d854d006aa6b… · 1.1 KB - translations/en-rodwell.jsonl
c5d200eb830cff6a… · 1.4 MB - translations/en-rodwell.LICENSE.md
f1a3bc6c8c857293… · 1.0 KB
How to cite this version
quranbench (2026). Quran corpus, version 0.8.0 [Data set]. Retrieved 2026-09-10 from https://quranbench.com/data#0.8.0. sha256(manifest.json)=00ca019ba8cf994001832177dabf43043494d8ba4c314f886879456891ed7a94.
Version 0.7.0
built 2026-07-26 · 77,881 tokens · 6,236 versesmanifest.json sha256
072bef82f79eaf47db61614ad2ee617eb35e220769db5831c19818ad6026298aThe complete artifact set for this version in a single archive: text, morphology, translations, numbering, mappings and the manifest. Because it embeds the GPL Leeds morphology (tokens.jsonl and morphology/), the tarball as a whole is GPL-2.0-or-later. Verify it against the published sha256 before use. For CC-BY-only text, take the per-file text group instead.
- quranbench-corpus-v0.7.0.tar.gz
aeb8574172c2fb68… · 6.5 MB
Uthmani text from the Tanzil Project, plus verse rows, surah metadata, numbering schemes, identifier mappings and the build manifest. Take this if you only want the Quran text under CC-BY. Full verse text lives in verses.jsonl.
- identifiers.json
ad691b421f7bcf70… · 1.8 KB - mapping/mapping.schema.json
26eeb699088f17ad… · 1.7 KB - mapping/v0.6.0-to-v0.7.0.json
27203bad598ab86d… · 535 B - numbering/kufan.json
43f70350303e536d… · 906 B - numbering/numbering.schema.json
31dfd3ef638e02b7… · 1.4 KB - sources.json
4be06a6070c84483… · 6.2 KB - surahs.json
b4f4992a245c4780… · 93.1 KB - verses.jsonl
987aa84729fc0890… · 5.4 MB
Root, lemma, part-of-speech and segment annotation from the Leeds Quranic Arabic Corpus. tokens.jsonl combines the Quran text with this annotation, so the combined file is GPL. If you cannot accept copyleft, use verses.jsonl from the text group instead.
- morphology/alignment-report.md
e99dee1e032fb2e1… · 2.6 KB - morphology/ATTRIBUTION.md
6582fc699af560b2… · 1.5 KB - morphology/LICENSE
edaef632cbb643e4… · 17.6 KB - morphology/roots.json
b482fea1fa5dafa6… · 2.2 MB - tokens.jsonl
28144efd420652fc… · 59.6 MB
Verse-level human translation editions in the public domain — the ones you may redistribute. Correspondence to the Arabic is verse-level only, never word-level. Each edition ships its own LICENSE file. Display-only editions (shown on the site under a licence that permits display but not redistribution, e.g. Talal Itani under CC BY-NC-ND) are deliberately excluded from these downloads and from the full tarball; they are listed below with the reason.
- translations/en-palmer.jsonl
7131a1c14f4ab442… · 1.3 MB - translations/en-palmer.LICENSE.md
82b68ac668616d5c… · 1.1 KB - translations/en-pickthall.jsonl
f84f0a5c532ae0e7… · 1.4 MB - translations/en-pickthall.LICENSE.md
3578d854d006aa6b… · 1.1 KB - translations/en-rodwell.jsonl
c5d200eb830cff6a… · 1.4 MB - translations/en-rodwell.LICENSE.md
f1a3bc6c8c857293… · 1.0 KB
How to cite this version
quranbench (2026). Quran corpus, version 0.7.0 [Data set]. Retrieved 2026-09-10 from https://quranbench.com/data#0.7.0. sha256(manifest.json)=072bef82f79eaf47db61614ad2ee617eb35e220769db5831c19818ad6026298a.
Version 0.6.0
built 2023-11-14 · 77,881 tokens · 6,236 versesmanifest.json sha256
aab85f59439ec917e441936a4d43d53908acd38d0bef4170f75848e7dc86b295The complete artifact set for this version in a single archive: text, morphology, translations, numbering, mappings and the manifest. Because it embeds the GPL Leeds morphology (tokens.jsonl and morphology/), the tarball as a whole is GPL-2.0-or-later. Verify it against the published sha256 before use. For CC-BY-only text, take the per-file text group instead.
- quranbench-corpus-v0.6.0.tar.gz
2b96fc3df465b91d… · 6.5 MB
Uthmani text from the Tanzil Project, plus verse rows, surah metadata, numbering schemes, identifier mappings and the build manifest. Take this if you only want the Quran text under CC-BY. Full verse text lives in verses.jsonl.
- identifiers.json
ad691b421f7bcf70… · 1.8 KB - mapping/mapping.schema.json
26eeb699088f17ad… · 1.7 KB - mapping/v0.3.0-to-v0.4.0.json
5c5bb838ed758ec8… · 503 B - mapping/v0.4.0-to-v0.5.0.json
f08ea8e7a62ab3a0… · 562 B - mapping/v0.5.0-to-v0.6.0.json
fccd06c2a124776d… · 561 B - numbering/kufan.json
43f70350303e536d… · 906 B - numbering/numbering.schema.json
31dfd3ef638e02b7… · 1.4 KB - sources.json
bc37730f27491860… · 4.8 KB - surahs.json
b4f4992a245c4780… · 93.1 KB - verses.jsonl
987aa84729fc0890… · 5.4 MB
Root, lemma, part-of-speech and segment annotation from the Leeds Quranic Arabic Corpus. tokens.jsonl combines the Quran text with this annotation, so the combined file is GPL. If you cannot accept copyleft, use verses.jsonl from the text group instead.
- morphology/alignment-report.md
e99dee1e032fb2e1… · 2.6 KB - morphology/ATTRIBUTION.md
6582fc699af560b2… · 1.5 KB - morphology/LICENSE
edaef632cbb643e4… · 17.6 KB - morphology/roots.json
b482fea1fa5dafa6… · 2.2 MB - tokens.jsonl
28144efd420652fc… · 59.6 MB
Verse-level human translation editions in the public domain — the ones you may redistribute. Correspondence to the Arabic is verse-level only, never word-level. Each edition ships its own LICENSE file. Display-only editions (shown on the site under a licence that permits display but not redistribution, e.g. Talal Itani under CC BY-NC-ND) are deliberately excluded from these downloads and from the full tarball; they are listed below with the reason.
- translations/en-palmer.jsonl
7131a1c14f4ab442… · 1.3 MB - translations/en-palmer.LICENSE.md
487cd57e90e17079… · 1.1 KB - translations/en-pickthall.jsonl
f84f0a5c532ae0e7… · 1.4 MB - translations/en-pickthall.LICENSE.md
3bcec95826428c18… · 1.1 KB - translations/en-rodwell.jsonl
c5d200eb830cff6a… · 1.4 MB - translations/en-rodwell.LICENSE.md
5a1d19775cbcb66e… · 1.0 KB
How to cite this version
quranbench (2026). Quran corpus, version 0.6.0 [Data set]. Retrieved 2026-09-10 from https://quranbench.com/data#0.6.0. sha256(manifest.json)=aab85f59439ec917e441936a4d43d53908acd38d0bef4170f75848e7dc86b295.
Display-only translations (not downloadable)
These editions are shown to readers on the site but are not part of the dataset: their licence permits display but forbids open redistribution, so they are excluded from every download above and from the full tarball. To use them, obtain the edition from its licensor under its own terms.
- Talal Itani (2012)CC BY-NC-ND 4.0licencedisplay-only
Verify it yourself
Every artifact’s sha256 is published here and inside the manifest. Download a file and confirm its checksum matches — no trust in this server required.
# 1. Download an artifact and the manifest for a version
curl -sO https://quranbench.com/api/v1/download/0.8.0/verses.jsonl
curl -s "https://quranbench.com/api/v1/manifest?version=0.8.0" > manifest-api.json
# 2. Hash the file you received
shasum -a 256 verses.jsonl
# 3. Compare it to the published checksum
node -e "const m=require('./manifest-api.json').manifest; console.log(m.checksums['verses.jsonl'].sha256)"
# The two hashes must be identical. Repeat for any artifact.