Skip to content

arXiv Bulk Data Access - Amazon S3

This page describes arXiv bulk data available from Amazon S3. See also details of other bulk data feeds from arXiv. Note that arXiv's S3 buckets are located in the Eastern US (N. Virginia) region.

Please review the Terms of Use for arXiv APIs before using the arXiv bulk data buckets.

Note: Most articles submitted to arXiv are submitted with the default arXiv license, which grants arXiv a perpetual, non-exclusive license to distribute the article, but does not assign copyright to arXiv, nor grant arXiv the right to grant any specific rights to others. We are thus unable to grant others the right to distribute arXiv articles. If you build indexes or tools based on the full-text, you must link back to arXiv for downloads. A small fraction of submissions are made with other licenses and this information is available in the OAI-PMH metadata.

Update 2023-03-23: Updates to reflect current total file sizes and mild restructuring of this page. Update 2011-01-03: Source files also available from S3, see below.
Update 2010-08-31: The arXiv PDF dataset has been updated and moved to a new bucket (arxiv). The locations below have been updated.
Update 2016-09-23: Tools section has been revised to reflect newer version of s3cmd.

A note on Regions and File Size

The complete set of processed arXiv PDF files and their source available from Amazon S3 in requester pays buckets (i.e. the downloader pays Amazon for the download based on bandwidth used, see Requester Pays Buckets in the Amazon S3 Guide. Please consult Amazon's aws S3 pricing page for their current rates for Data Retrival, available at Amazon S3 pricing.) Our use of Amazon requester pays means that we can open downloads to anyone with predictable cost, and avoids putting any additional load on our servers that might impact interactive performance. Note that arXiv's buckets are located in the Eastern US (N. Virginia) region.

The complete set of files as of March 2023 is about 5.6 TB. With an estimated growth rate of around 100 GB per month, we expect increased growth as the submission rate continues to increase over time. The PDF files are about 2.7 TB, with the source files making up the remaining 2.9 TB of the estimate.

Bulk PDF Access

PDFs are available on S3 in the arxiv requester pays bucket. They are grouped into .tar files of ~500MB each (which we've found is a good size chunk). The complete set of PDF as of March 2023 is about 2.7 TB, with an estimated growth rate of around 100 GB per month, which we expect will increase as the submission rate increases over time. Examples keys for these files with the arxiv bucket are: Examples keys for these files with the arxiv bucket are:

pdf/arXiv_pdf_1001_001.tar         (s3://arxiv/pdf/arXiv_pdf_1001_001.tar in s3cmd URI style)
pdf/arXiv_pdf_1001_002.tar         (s3://arxiv/pdf/arXiv_pdf_1001_002.tar)
pdf/arXiv_pdf_1001_003.tar         (s3://arxiv/pdf/arXiv_pdf_1001_003.tar)

Which are chunks 1, 2 and 3 for month 1001 (2010-01). The complete list of all chunks if provided in a manifest with some additional information including dates and checksums. The manifest is:

pdf/arXiv_pdf_manifest.xml         (s3://arxiv/pdf/arXiv_pdf_manifest.xml)

and has the following format:

<?xml version='1.0' standalone='yes'?>
<arXivPDF>
  <file>
    <content_md5sum>1852974c8570cdafd91522ee93719ee5</content_md5sum>
    <filename>pdf/arXiv_pdf_0001_001.tar</filename>
    <first_item>astro-ph0001001</first_item>
    <last_item>hep-th0001208</last_item>
    <md5sum>4b5eeb603fd68bb05b9dd3341e9067fb</md5sum>
    <num_items>1751</num_items>
    <seq_num>1</seq_num>
    <size>526080000</size>
    <timestamp>2009-12-23 14:41:24</timestamp>
    <yymm>0001</yymm>
  </file>
  <file>
    <content_md5sum>650da80f3bcd1f4cd3d994b572ecdbb9</content_md5sum>
    <filename>pdf/arXiv_pdf_0001_002.tar</filename>
    <first_item>hep-th0001209</first_item>
    <last_item>quant-ph0001119</last_item>
    <md5sum>eedc2d7c09cf11fda188d8600c966104</md5sum>
    <num_items>565</num_items>
    <seq_num>2</seq_num>
    <size>139560960</size>
    <timestamp>2009-12-23 14:42:52</timestamp>
    <yymm>0001</yymm>
  </file>
  ...
</arXivPDF>

where there is one <file> for each chunk file. The elements are:

content_md5sum : MD5 sum of all the files in the tar package concatenated but not packaged. Use md5sum for the md5sum of the tar package which should match the S3 MD5 sum.

filename : Name of file within bucket, prefix bucket name s3://arxiv/ for complete identifier

first_item and last_item : arXiv identifier of article PDF first in tar package, and last in tar package

md5sum : MD5 sum of tar package, can be used as check against downloaded file

num_items : Number of PDF files in tar package

seq_num : Sequence number within month yymm

size : Size of tar package in bytes

timestamp : Timestamp of tar package (unix mtime when created, expressed at YYYY-MM-DD HH:MM::SS)

yymm : Two digit year and month of items in the tar package. Starts with 9108 for 1991-08, rolls past y2k to 0001 for 2000-01, 1008 for 2010-08 etc.

Bulk Source File Access

Similar to the processed PDF files, the arXiv source files (mostly TeX/LaTeX with figures in tar.gz format) available from Amazon S3 in requester pays buckets.

The source files are available on S3 in the arxiv requester pays bucket using an arrangement similar to the PDF files described above. They are grouped into .tar files of ~500MB each and the complete set of source files is about 2.9 TB (March 2023). Examples are:

src/arXiv_src_1001_001.tar         (s3://arxiv/src/arXiv_src_1001_001.tar in s3cmd URI style)
src/arXiv_src_1001_002.tar         (s3://arxiv/src/arXiv_src_1001_002.tar)
src/arXiv_src_1001_003.tar         (s3://arxiv/src/arXiv_src_1001_003.tar)

Which are chunks 1, 2 and 3 for month 1001 (2010-01). The complete list of all chunks if provided in a manifest with some additional information including dates and checksums. The manifest is:

src/arXiv_src_manifest.xml         (s3://arxiv/src/arXiv_src_manifest.xml)

See notes above for the format of this manifest which has root element arXivSRC.

Update frequency

We update both the bulk PDFs and source files to add new content on an approximately monthly schedule. Updates to existing files are less frequent. Please contact arXiv administrators if you propose to build a service relying upon a particular update schedule.

Tools

We do not track development of tools interacting with Amazon S3, nor endorse any particular tool. However, in development of this facility on a Linux platform we have found s3cmd useful. Please consult your tool's documentation for the appropriate syntax and usage.