We released Seed 1.6. I was responsible for the multimodal mixed continual training (MMCT) stage, enabling the model to achieve strong native multimodal capabilities without sacrificing its text performance.