Awesome Mixture-of-Experts (MoE) Papers and Source Codes

HunyuanImage-3.0: The Largest Open-Source Multimodal Image Generator with Native Reasoning and MoE Architecture 2562

HunyuanImage-3.0 is a groundbreaking open-source image generation model developed by Tencent. Unlike traditional diffusion-based approaches, it builds a native multimodal…

12/26/2025Mixture-of-Experts (MoE), Multimodal Reasoning, Text-to-Image Generation