Together AI

Together AI has announced a $102.5M Series A funding round, highlighting their cloud platform for generative AI. The platform, built on cutting-edge AI research, offers the fastest inference stack through API calls. Noteworthy innovations from their research team include Cocktail SGD, FlashAttention 2, and sub-quadratic model architectures like Monarch Mixer. Monarch Mixer, developed in collaboration with Hazy Research, introduces an efficient model architecture, initially targeting BERT with 25% parameter efficiency improvement and 3x faster training. Together AI also supports open-source initiatives like RedPajama, offering custom models, GPU clusters, and an end-to-end AI platform for building, deploying, and scaling applications.