On September 1, 2026, Light Origins made a groundbreaking move by officially releasing and open-sourcing its universal navigation brain, LightNav-0. This innovative system harnesses the power of the Real2Sim2Real technology and has been trained on an extensive dataset comprising over 2,000 real-world indoor and outdoor scenes. Through this, it has synthesized an impressive 4,000+ hours of post-training data on a large scale. As a result, LightNav-0 demonstrates exceptional cross-agent, cross-task, and cross-scenario zero-shot navigation capabilities in real-world settings, eliminating the need for further data collection or model fine-tuning. LightNav-0 has established itself as a leader across public visual language navigation (VLN) evaluation benchmarks, marking a pivotal achievement in the scaling alignment phase of Light Origins' three-stage approach: 'scaling pre-training—scaling alignment—scaling deployment.'