We are a generative AI startup. Leveraging breakthrough AI research, we have developed a platform for training next-generation large language models powered by diffusion. Unlike existing auto-regressive models, which only output one token at a time, our models can output many tokens in parallel. Founded by Stanford CS faculty.