Tencent has released and open-sourced HunyuanImage 3.0, an 80-billion-parameter native multimodal image generation model. The company says it is the first open industrial-grade model of its kind, with performance comparable to leading non-open-source models. The model can leverage knowledge for reasoning, parse instructions exceeding 1,000 characters, and render long text strings in generated images. It follows HunyuanImage 2.0, introduced in May, which offered millisecond-level response, photorealistic quality, and real-time typing-to-image output. [TechNode reporting]
Related
