Academic Torrents

academictorrents.com · Storage · 1 record · JSON

Academic Torrents (tracker + index, with optional donor mirroring) Storage

caveatinstitutionno cardRper-object-cap

academictorrents.com · academic · storage/academic-torrents · JSON · compare

Free tier	NO BYTE QUOTA ON THE INDEX, because the index is all they host: 'It's important to note that we don't directly host your data. However, we manage a network of donated servers. When you upload a torrent file to our site, hosting nodes (seeds) register with the tracker to provide download locations for the data.' The catalogue is 419.44 TB across the network but that is other people's disk. The only hard number is on the torrent file itself: a 10 MB per-torrent limit, stated by the maintainer ('each of them is beneath the 10MB size limit') and defended against a request to raise it. unit TB; metered on unstated
Price	none stated
Per-object cap	per-object-cap: 10 MB (10000000 bytes)
Retention	per-object-cap: 10 MB for the .torrent file only; no cap on the dataset it describes. retention: entries are permanent in practice; the PDF describes a network designed for durability across node failure
Egress	Free, and structurally so - it is BitTorrent, so egress is other peers' bandwidth. Also offers an HTTP fallback: 'Backup URLs (also called WebSeeds)' are injected into the .torrent so the swarm can be skipped entirely. Their own docs note most free hosts (Google Drive, Dropbox) do NOT work as webseeds because they lack RANGE support, while archive.org and Zenodo (single-file only) do.
Identity	institution derived from the record's prose, not from an identity field, and no signup was executed: y across node failure. request-rate: not documented on the API page. There is no seed guarantee - 'as long as at least one person can become a seed for that data' - so durability is
Limits	per-object-cap per-object-cap: 10 MB source per-object-cap: 10 MB for the .torrent file only; no cap on the dataset it describes. retention: entries are permanent in practice; the PDF describes a network designed for durability across node failure. request-rate: not documented on the API page. There is no seed guarantee - 'as long as at least one person can become a seed for that data' - so durability is a function of who else mirrors it.
Binding	per-object-cap the per-object cap sets the shard size and therefore the object count; it bounds the total more tightly than any headline quota
Interface	rest rest
Tooling	REST v2 at https://academictorrents.com/apiv2/ - POST /apiv2/entry with `file` as a base64-encoded .torrent and `urllist` as the webseed URLs; API key required. Also POST /apiv2/collection. A first-party uploader exists (github.com/academictorrents/academictorrents_uploader) and a `mirroring.md` doc for running a Smart Node that seeds selected collections from an RSS feed. Standard BitTorrent clients (transmission-daemon recommended) do the actual transfer.
Composes with	This is a discovery and swarm-coordination layer, not storage. What it multiplies is availability: one .torrent entry (10 MB) makes an arbitrarily large dataset reachable through unlimited peers, and the webseed lists let you point at storage you already have elsewhere (archive.org items, Zenodo records, your own HTTP server). Chain shape: bytes live on IA/Zenodo/your server -> webseed URL -> AT entry -> swarm plus HTTP fallback. Verified that the mechanism is real: a fetched .torrent carries `url-list` with https://files.inria.fr/ and the AT announce URL.
Reliability	The fatal caveat: without a donor node or a webseed, an entry is a pointer to nothing. Their model is explicitly 'the burden of hosting data' distributed to volunteers, so AT gives zero storage guarantee. It is also not a place to put bytes - upload.php returned 409 and there is no bulk endpoint. Treat it as a distribution amplifier for data that already has a home, never as a home.
Notes	NO BYTE QUOTA ON THE INDEX, because the index is all they host: 'It's important to note that we don't directly host your data. However, we manage a network of donated servers. When you upload a torrent file to our site, hosting nodes (seeds) register with the tracker to provide download locations for the data.' The catalogue is 419.44 TB across the network but that is other people's disk. The only hard number is on the torrent file itself: a 10 MB per-torrent limit, stated by the maintainer ('each of them is beneath the 10MB size limit') and defended against a request to raise it.
Evidence rung	reported One page read moves it to documented.
Checked	2026-10-21 (inherited); stale on 2026-11-04
Parsed numbers	`$.limits[0].value` limit:per-object-cap 10.0 MB

Evidence (6)

date	source	what it proved
2026-10-21	academictorrents.com/docs/uploading.html	'we don't directly host your data. However, we manage a network of donated servers... Alternatively, you can use a Backup URL... where the data will be mirrored.'; WebSeeds section; RANGE queries section
2026-10-21	academictorrents.com/	'Making over 419.44TB of research data available!'
2026-10-21	github.com/academictorrents/academictorrents-docs/issues/46	maintainer ieee8023: 'I'm against raising the per file limit because that's a lot of space to take up for just a torrent file'; reporter confirms 'each of them is beneath the 10MB size limit'
2026-10-21	academictorrents.com/download/cf445f6073540af0803ee345f46294	probe: GET -> 200, application/x-bittorrent, 100,785 bytes; bencoded body contains 'url-list' with https://files.inria.fr/ and announce https://academictorrents.com/announce.php
2026-10-21	academictorrents.com/docs/api.html	POST /apiv2/entry with base64 torrent + urllist, API key required
2026-10-21	academictorrents.com/upload.php	probe: GET -> 409 (no anonymous bulk write path)

Unknown or withheld values (3)

$.free_tier.notes process
$.per_object_cap.notes not-extracted
$.retention.notes not-extracted

Built from commit 872967ae41, data as of 2026-10-27. Every value links to its source; freshness badges are computed in your browser. How to read this.