Comfy-Org released an adapted version of the Qwen-Image-2.1 model

The Comfy-Org team presented Qwen-Image-2.1 — a specialized build of the popular generative model, optimized for operation in the ComfyUI environment. The release allows users to integrate image generation capabilities directly into node workflows without the need to use external APIs.

For the correct operation of the system, strict adherence to the folder structure is required. Users must place the model components, including diffusion files, text encoders, and VAE, in the corresponding directories within the ComfyUI structure. The model demonstrates a result of 89 points with a generation time of 1 hour on standard configurations.

The release includes ready-made workflow templates designed to perform text-to-image and image editing tasks. The toolkit is oriented towards local deployment, which provides full control over the generation process and eliminates the limitations characteristic of cloud services.

Adaptation to ComfyUI simplifies the introduction of modern visual models into automation pipelines. The use of local versions allows developers and enthusiasts to experiment with generation parameters using the familiar node interface, which significantly lowers the entry threshold for working with the Qwen architecture in creative tasks.