Modernizing arXiv technology: Updates on the arXiv-CE project

This past August, arXiv turned 35 years old. That’s 35 years of open science – and 35 years of legacy tech.

arXiv is one of the largest and most widely used open repositories in the world. It’s also one of the oldest – and that comes with its share of technical debt. As arXiv has expanded and submissions have increased, aging infrastructure and growing requirements have made it clear that arXiv’s tech stack needed to be modernized so that we can scale sustainably and continue to serve the research community.

In 2022, arXiv proposed a plan to modernize our technology and migrate arxiv.org to the cloud – the arXiv-CE (Cloud Edition) project. Since 2023, the arXiv team has been diligently working to complete this plan, which includes updating our code base and migrating off of Cornell IT servers.

Thanks to generous support from major funders, arXiv has been able to hire new members to our engineering team who dedicate their talents specifically to the arXiv-CE project. arXiv has also recently welcomed our inaugural Engineering Director, Jeff Sank, who hit the ground running in April of this year. With over 20 years of experience, Jeff’s guidance and leadership as we near the completion of this critical project have been invaluable to arXiv.

The tech goals for the arXiv-CE project include improving service for all users, securing arXiv and maintaining stability for years to come, and providing updated features for researchers using arxiv.org. Meeting these goals – while ensuring minimal disruptions to service – is a huge task. arXiv’s engineering team has already made marked progress on these goals that support arXiv submitters and readers.

Other arXiv-CE completed updates include our update to Submission 1.5, reducing delays in new submission announcements, and steady improvement in arXiv’s Experimental HTML, which makes research more accessible to researchers with disabilities. In August 2026, arXiv made its entire collection available in HTML: 4.6 million article versions, with 99% of all articles submitted as LaTeX now readable in HTML. arXiv browse has already been completely migrated to the cloud as of August 2023. As we continue to make progress on arXiv-CE, arxiv.org has also seen fewer outages, despite more traffic than ever. From October 2025 to October 2026, arXiv had no outages at all, with 99.96% availability.

As of October 2026, the arXiv-CE project is very close to the finish line. There are still many critical updates to complete, including launching Submission 2.0 and migrating arXiv’s database. We anticipate some of these updates will require scheduled maintenance on arxiv.org and potentially disrupt submission service and announcements. arXiv is a critical service to researchers who visit arxiv.org every day, and our goal is to keep disruptions as minimal as possible. All outages will be announced in advance so that arXiv users can plan their submissions and other work on arxiv.org as needed.

Thank you to our funders and the scientific community at large who support arXiv and make these massive undertakings and important projects possible. And thank you to Cornell University for 25 years of collaboration, support, and hosting arxiv.org on Cornell IT servers. The completion of arXiv-CE will make arXiv’s technology more secure, flexible, and truly open, independent, and modern.