Alibaba's Tongyi Qianwen team has open-sourced Qwen-Image-2.1-Turbo, an accelerated image generation model capable of producing and editing images in just eight denoising steps. Built on a 7-billion-parameter architecture, the model supports 2K resolution output and continuous natural-language editing. Model weights are now available on Hugging Face and ModelScope, while Pro and Turbo APIs have also gone live for immediate developer access.