Photoplethysmography (PPG)-based foundation models are gaining traction due to the widespread use of PPG in biosignal monitoring and their potential to track diverse health indicators. In this paper, we introduce Pulse-PPG, an open-source PPG foundation model trained exclusively on raw PPG data collected over a 100-day field study with 120 participants. Existing open-source PPG foundation models are trained on clinical data, and those trained on field data are closed source, limiting their applicability in real-world settings. Extensive evaluations demonstrate that Pulse-PPG, trained on uncurated field data, exhibits superior generalization and performance across clinical and mobile health applications in both lab and field settings, when compared with state-of-the-art PPG foundation models trained on clinical data. Exposure to real-world variability in field-collected PPG data enables Pulse-PPG to learn more robust representations. Furthermore, pre-training Pulse-PPG on field data outperforms its own pre-training on clinical data in many tasks, reinforcing the importance of training on real-world datasets. To encourage further advancements in robust PPG modeling, we have open-sourced*our Pulse-PPG model, providing researchers with a valuable resource for developing the next generation of task-specific PPG-based models.