We present the first release of p-HCP (Prenatal Human Connectome Patterns), an imaging dataset of human fetal brain development covering the second half of gestation. This dataset was acquired ex vivo using magnetic resonance imaging (MRI) at ultra high field (11.7~teslas), and includes whole-hemisphere T2-weighted images at 100 {micro}m isotropic resolution, quantitative relaxometry (T1, T2, and T2*), and high angular resolution diffusion-weighted images for multiple b-values at 200 {micro}m. Brains larger than the workspace of the small-bore scanner were sectioned into blocks, acquired blockwise, and digitally reconstructed using a dedicated semi-automatic method. This initial data release includes three gestational ages (18, 27, and 31 post-conceptional weeks) with a complete set of anatomical images, relaxometry maps, and diffusion-based microstructure measurements. This dataset offers new opportunities to investigate neurodevelopmental processes that have not yet been explored with full three-dimensional coverage at this resolution by MRI, and may serve as a multimodal mesoscopic reference template for the fetal brain.