First community-wide machine learning gravitational wave search challenge with 6 algorithms tested on progressively realistic data including real O3a noise. Top ML methods achieve 95% of matched filtering sensitivity for Gaussian noise and 70% for real noise.