DeepSeek-V4-Flash-Vision-Exp multimodal model is now available in the DeepSeek adapter, enabling visual understanding capabilities for image analysis tasks.