X has open-sourced its recommendation algorithm, which powers the "For You" feed, after nearly three years. The algorithm combines content from followed accounts and machine learning-discovered posts, ranking them using a Grok-based Transformer model. This model predicts user behavior probabilities, such as likes and shares, to score and recommend content. The open-source release aims to enhance transparency and address criticisms of algorithmic bias. Despite the open-source release, some components, like weight parameters and training data, remain undisclosed. Users have noted that while the architecture is available, the absence of these elements limits full transparency. The move is part of a broader effort by Musk to make X's systems more transparent, following previous controversies over content distribution biases.