The Qwen AI platform has officially launched the Qwen-Image-2.1 Pro and Turbo APIs. This integration allows developers to incorporate image generation and editing capabilities directly into their applications without the need for independent model deployment or infrastructure management.
The Qwen-Image-2.1-Turbo model is built on a 7B visual generation architecture. It optimizes the denoising process for image generation and editing to 8 steps while maintaining support for 2K resolution output. The model weights for Qwen-Image-2.1-Turbo have also been released.
The updated models support the input of up to 10 reference images. Users can define specific editing areas using selection, brushing, or independent masks. The 2.1 version improves consistency in portrait and product editing, specifically preserving facial identity features and maintaining text, texture, and shape integrity.
Developers can utilize these tools for various workflows, including e-commerce product material creation, marketing poster design, and fashion styling. The API enables users to perform image generation and editing tasks within their existing interfaces, streamlining the transition from text concepts to visual assets.
