Together AI, a startup focused on enhancing the efficiency of pre-training and fine-tuning open source foundation models on high-performance graphics chips, has successfully closed a significant $102.5 million Series A funding round. The round is led by Kleiner Perkins, with participation from Nvidia and Emergence Capital.
Founded by serial entrepreneur and former Apple executive Vipul Ved Prakash, Together AI has assembled a team of seasoned artificial intelligence researchers. One notable member is Tri Dao, the company’s chief scientist, who created FlashAttention, an open-source library crucial for training foundation models more efficiently.
The startup’s cloud platform serves as a comprehensive solution for open source AI, enabling companies to harness the capabilities of advanced graphics chips like Nvidia H100s. The platform aims to streamline the pre-training and fine-tuning processes, making them more cost-effective and accessible.
Several notable companies, including Pika Labs, Nexusflow, Voyage AI, and Cartesia, leverage Together AI’s cloud platform. Additionally, the startup collaborates with partners such as Crusoe (focused on using stranded energy to power AI chips), Nvidia, and cloud infrastructure provider Vultr.
The substantial Series A funding positions Together AI as a key player in the open source landscape, challenging established companies like OpenAI and Anthropic. The support from Kleiner Perkins, Nvidia, and Emergence Capital underscores the promising potential of Together AI’s approach to optimizing open source AI workflows.
