Intra-surgical optical coherence tomography (OCT) complements surgical microscopes by adding a third spatial dimension to the imaging context during ophthalmic interventions. Recent advancements in OCT technology and artificial intelligence are driving innovations in computer-guided and robotic ophthalmic surgery. However, developing and evaluating models for OCT reconstruction, visualization, and automated scene understanding demands suitable datasets and annotations. For this purpose, we present the porcine anterior segment OCT (PASO) dataset. It contains 141 volume scans acquired from 47 enucleated porcine eyes using a cutting-edge microscope-integrated swept-source OCT prototype with a field of view of 11.8 mm by 11.8 mm and an imaging depth of around 5 mm. The anterior segments, including cornea, iris, and anterior lens surface, were imaged before, during, and after tissue manipulation with 12 microsurgical instruments, resulting in a multi-faceted dataset. The dataset contains raw spectra as well as reconstructed scans averaged from 12 repetitions. Additionally, for 1020 cross-sectional scans extracted from 19 porcine eye volumes, surgical instrument segmentation (SIS) masks are provided, generated using the segmentation foundation model SAM and manually refined. Baseline algorithms are presented for reconstruction and adjustable averaging of scans, and for surgical instrument segmentation. As the first publication of raw and reconstructed OCT scans of ex-vivo porcine eyes during instrument-tissue manipulation, our dataset serves as a valuable catalyst for progress in the field of computer- and robot-assisted ophthalmic surgery research.