Qwen-Image-2512 is a state-of-the-art open-source text-to-image (T2I) model that delivers exceptional realism in image generation. As part of a family of leading speech models (available in 0.6B and 1.7B parameter sizes), it supports 10 languages, addressing the need for versatile and high-quality cross-lingual visual content creation. Key highlights include prompt-based Voice Design for customized audio-visual integration, 3-second zero-shot voice cloning for quick personalization, and extreme low-latency streaming for seamless real-time applications, making it ideal for diverse creative and practical use cases.