ULTERA Database, developed since May 2021 under the ARPA-E's ULTIMATE program Phase 1 and 2, is the largest collection of literature data on High Entropy Alloys (HEAs), and more broadly Compositionally Complex Alloys / Materials (CCA/CCMs) or multi-principle-element alloys (MPEAs), designed from the ground up to enable rapid ML-based discovery of novel materials using forward and inverse design techniques. Our primary focus is to build a robust and reliable community resource by collecting a wide spectrum of chemical elements (see below), 50+ properties (mechanical, magnetic, thermodyanmic, and other thermophysical properties), and 45+ processing techniques. As of May 2025 / Alpha Release, ULTERA Dataset contains over: 12,540+ manually-extracted experimental property datapoints for HEA/MEAs (or 15,462 including lower-entropy support data), corresponding to 4,270+ unique experimentally observed HEA/MEAs materials, collected from 813 individual literature publications / DOIs, reporting on 830 chemical systems formed by combinations of 59 chemical elements, out of which 21 were present in over 100 unique materials and 37 were present in over 20 unique materials. All of the data in ULTERA is heavily processed through thousands of lines of code enabling us to integrate starting literature data with decisions on next experiments and improvments in ML pipelines. Homogenize data and detect highly curated with many steps of data validation and then processed through our abnormal data detection tools (pyqalloy.ultera.org). The above data count numbers would be noticably higher, if not for our "aggresive" duplicate detection catching not only near-exact duplicates, but also most of hidden duplicates in the literature. For instance, ULTERA will retain original study data point reporting that "alloy A2BC2D was arc-melted, die cast, and homogenized resulting in hardness of 2.32GPa", and remove a review study data point reporting on "alloy A33.3B16.6C33.3D16.6 had hardness of 2.3GPa after annealing" even if no correct reference was given. All data is available through a high-performance API, designed to follow FAIR principles upon public release, while this Zenodo repository serves as an archive and reference for ML studies using specific version of ULTERA. Detailed statistics and addition infromation can be found at our ultera.org project web page. The main scope of this dataset is collecting data on compositionally complex alloys (CCAs), also known as high entropy alloys (HEAs) and , with extra attention given to (1) high-temperature (refractory) mechanical data, (2) phases present under different processing conditions. Although low-entropy alloys (incl. binaries) are typically not presented to the end-user (or counted in statistics), some are present and used in ML efforts; thus, all high-quality alloy data contributions are welcome! You can set up a contribution in as little as few minutes with this contribution repository at contribute.ultera.org