The Windbench portal for wind energy verification and validation (V&V) repositories is being redesigned, using an Open Science approach, with the objective of improving data openness and reproducibility of research. Windbench manages V&V benchmarking by allowing researchers to meet online and share their data for model intercomparison. The portal follows the IEA Task 31 "Wakebench" model evaluation protocol 2 for wind farm flow modelling based on a building-block validation approach that allows to systematically evaluate the wind farm multi-scale system by decomposing it in a number of sub-system and unitary building blocks of increasing complexity. The new portal integrates Jupyter notebooks in the cloud to allow researchers run their scripts for post-processing and model evaluation using their own data or data shared by other benchmark participants. These notebooks allow researchers to create and share documents that contain text, equations and visualization of results together with the code that generates these results. This improves reproducibility and provides better means for the community to adopt model evaluation standards by reusing contributions from previous benchmarks in terms of, for instance, data filtering and post-processing scripts, error metrics, visualization tools, etc. The GABLS3 diurnal cycle benchmark 3,4 for mesoscale to microscale atmospheric boundary layer models is used as case study to demonstrate the benefits of the open science approach in the Wakebench V&V framework. The benchmark has been organized within the New European Wind Atlas (NEWA) project to help microscale model developers implement solutions for introducing mesoscale forcings (that could be extracted from a wind atlas). The case is also used to explain how V&V benchmarks contribute to open source community models like CFDWind 3.0, based on OpenFOAM and developed for the NEWA model-chain.