Tencent has released Hunyuan Image 3.0, an open-source multimodal image generation model with 80 billion parameters.

The company said the model is designed for industrial use, and claims its performance is on par with leading closed-source alternatives.

Hunyuan Image 3.0 can interpret complex semantics, generate images, and produce lengthy text outputs.

The launch follows the earlier release of Hunyuan Image 2.0 in May, which focused on faster response times and more realistic images.

Tencent has also developed related tools, including 3D generation models, a customizable image generation plugin called InstantCharacter, and a multimodal video generation tool named HunyuanCustom.

🔗 Source: AI Base

Recent Tencent developments