Skip to content
United States of America
File

ProtoStar: open scientific data for AI

This dossier tracks ProtoStar’s reuse of public scientific datasets, models and code and challenges testing its value.

Last updated : 1 key moments

Where things stand

As of 29 October 2026, the initiative is announced with a first climate challenge planned for 2–13 November; results remain pending.

Hugging Face and universities launch ProtoStar for AI-ready open science

Hugging Face announced Project ProtoStar on 29 October 2026 with universities and open-science partners. It aims to help publish publicly funded datasets, models and code in forms researchers and AI systems can reuse.

Hugging Face logo identifying ProtoStar’s infrastructure partner.
Hugging Face logo identifying ProtoStar’s infrastructure partner. — Victor / Hugging Face / Wikimedia Commons — emoji Apache 2.0 https://www.apache.org/licenses/LICENSE-2.0 ; text public domain ; PNG rendering.

Partners include Carnegie Mellon, Purdue, Georgia Tech, Arizona State and George Washington. Participation uses non-binding, non-exclusive commitments, with institutions retaining control over which resources to release.

A Carnegie Mellon climate challenge is planned for 2–13 November using data from 210 US cities and comparisons with simple statistical methods. The launch post and project page differ on the final forecast year, 2026 or 2025; participants need to check the rules. The challenge has not yet established a scientific improvement.

Sources 2