Tencent releases and open sources "Mixed Image 3.0"

On September 28th, Tencent Hydrix launched and open-sourced the native multimodal image generation model "Hydrix Image 3.0", with a parameter scale of 80B. It is reported that this model is the first open-source industrial-grade native multimodal image generation model, and Tencent has stated that the model's performance can be compared to top closed-source models in the industry. In addition, the model can utilize knowledge for reasoning, analyze complex semantics at the thousand-word level, and generate long-form text.