We introduce the first learning-augmented streaming algorithms for Correlation Clustering on both complete and general graphs, improving the best-known space-approximation tradeoffs.