Ai2 and Hugging Face announced an expanded partnership to make fully open AI models and datasets easier to find, customize, and build on. Under the new agreement, Hugging Face is roughly tripling Ai2's storage there to nearly two petabytes, effective immediately, and Ai2's traffic is no longer subject to the Hub's standard rate limits, allowing even the largest datasets and multi-checkpoint models to download at full throughput. Ai2 publishes training data, intermediate checkpoints, data mixes, evaluations, ablations, and demos to support open science, and according to Hugging Face's official heatmap, Ai2 publishes more new artifacts each year than any other organization it tracks. Hugging Face is also a collaborator on Ai2's open science, having wired olmOCR-Bench into the Hub's leaderboard workflow and bringing MolmoAct 2 into LeRobot, its open-source robotics library, with training data released in LeRobot's standard format. Benchmarks like IFBench, RewardBench 2, and AstaBench live on the Hub, alongside OlmoEarth models and climate models like HiRO-ACE, part of Hugging Face's Hugging Science collection. Additionally, Asta's literature search tool now runs Deep search by default, and an "Explore with Asta" click opens AutoDiscovery hypotheses in DataVoyager. A new engineering write-up covers infrastructure for open Earth observation models, which can run inference across a continent in roughly a day at fractions