Yesterday.TencentIt's officialOpen SourceNativeMulti-modular biograph model「Mixed Image 3.0 (HunyuanImage 3.0)" with a high of 80B。
It was described as the first open-source industrial-grade multiple-modular model, as well as the current open-source model with the strongest effect and the largest amount of parameters, with effect on the head closed-source model of the target industry。
Mixtures image 3.0 has been significantly enhanced in semantic understanding, aesthetic sense and reasoning, capable of deciphering a thousand-word complex semantics and generating high-quality images。
Unlike the traditional multi-model combination, hybrid image 3.0 uses a native multi-modular structure that allows the completion of multi-modular input outputs such as text, pictures, video and audio within a single model。
Officially, the model has not only the ability to paint, but also the common sense and reasoning of linguistic models. For example, the entry of a hint "to produce a full-eat four-gram caricatures of one month's duration" allows the model to generate a full-size comic without a case-by-case description。
In addition, Mixed Image 3.0 is prominent in text generation, complex poster design, cartoon illustrations, etc., and is able to meet the diverse needs of illustrators, designers and content creators and significantly increase the efficiency of creation。
The current open version only supports the graphic functions, and the graphics, image editing, multi-wheel interaction, etc., will be gradually rolled over。
Users can experience the model via computer-based access to the Mixer Network (https://hunyuan.tencent.com/image), and the model weights and acceleration versions are on line with open-source communities such as Github, Hugging Face, and can be downloaded free of charge by businesses and individual developers。
Github: https://github.com/Tencent-Hunyuan/HunyuanImage-3.0
Hugging Face: https://huggingface.co/tencent/HunyuanImage-3.0