Native multimodal coding from images, videos, design drafts and document layouts. Leading vision coding, tool use, and GUI agent performance. From 智谱 / Z.ai.