MeissonFlow Research

We are thrilled to announce the official establishment of MeissonFlow Research, a non-profit research group.  As steadfast explorers in the field of AI research, our mission is to develop and lead the global for front of generative models, widely applying them across various domains such as design and media. We firmly believe that efficient, controllable, and high-quality generative models will significantly expand the boundaries of human creativity and enrich the diversity of design. We are convinced that generative AI will become a cornerstone technology driving human societal progress. By enabling a broad user base to easily run these high-quality models on personal computers, we will allow more people to benefit from the most advanced AI achievements. 

Our vision is to establish a global leading exemplar in generative models, spearheading the vigorous development of this field. As a starting point to achieve this grand objective, we are proud to unveil our first model.  Today, as the first step towards this goal, we release Meissonic-v1.0  that push the frontiers of Non-Autoregressive Transformers for Text-to-Image synthesis.