# Vision-Language Understanding and Generation

BAGEL is an open-source, 7-billion-parameter active multimodal foundation model trained on large-scale interleaved multimodal data, excelling in both understanding and generation tasks.