Alibaba’s Kevin team released a new image generation artificial intelligence (AI) model last week. Dodged Kevin VLO, this is the successor of the Queen 2.5 Vision Language model and comes with several upgrades than the old models. The latest AI image model supports both image and image -to -image generation from text. It also supports text input in several languages, including English and Chinese. In addition to the image generation, the AI model is also capable of editing the input images as well as input images.
Kevin VOL accepts indication in multiple languages
In a post on X (known as Twitter), the official handle of the Kevin team announced the release of a new model. The technical name of the model is QWEN3-235B-A22B, and it is available for free here on the company’s chat interface. Users can use the model without logging in.
Members of the Gadget 360 staff tested the AI model and found it to be the image generation to be equal to Google’s Imagine 2. The quality of the direction and image output is slightly lower than the feature of Imagin -3 and Openi’s GPT -4O -strength image generation. However, the time of his generation is faster than both of them, and the rate is higher.
On its gut hub page, the company said that the Kevin VLO comes with a better understanding of iconic, which enables better in -line modification without distorting the structural integrity of the input image. It also improves the overall quality of the output. The model also better understands the ambiguous and open indicators, and can create images that are associated with user expectations.
In addition to image generation and editing, Canvin VLO can perform the imagery -related tasks such as edge detection, section, forecasting maping, and more. The company said the future version of the model would also be able to accept multiple input images and connect them based on user requests.
The latest AI image generator has also been improved with text rendering. We were able to produce the correct text in various fonts in the model test. Finally, the Kevin VO also supports images as input with the dynamic aspect ratio, which includes extreme proportions such as 4: 1 and 1: 3. The company plans to add the feature to produce images in proportion to various aspects.
