Together AI, a generative AI startup that provides a cloud platform focused on open source models, announced today that it has secured $102.5 million in Series A financing. The substantial round was led by Kleiner Perkins, with additional participation from NVIDIA, Emergence Capital, and over a dozen other investors.
The new capital will power Together AI’s goal of providing developers everywhere access to a fast, reliable cloud platform for building and deploying generative AI applications. Unlike other vendor-controlled solutions, Together AI allows its users to pre-train, fine-tune and run leading open source models as well as proprietary models they create and own.
As AI rapidly moves from research labs into real-world software, Together AI aims to ensure generative models don’t become walled gardens controlled by one or two dominant players. Its open source focus provides flexibility for enterprises to incorporate AI without fear of vendor lock-in.
“Startups and enterprises alike are looking to build a generative AI strategy for their business that is free from lock-in to a single vendor,” said CEO Vipul Ved Prakash. “Open source AI provides a strong foundation for these applications with increasingly powerful generative models being released almost weekly."
Prakash highlights Together AI's existing traction and growth, stating that "Since launch in June 2023, our AI compute, training, and inference products have seen tremendous adoption from startups and enterprises, and we plan to build on this momentum by scaling our services and introducing new products."
Together AI differentiates through research innovation and infrastructure optimization. On the research side, initiatives like its large-scale RedPajama-V2 open dataset demonstrate commitment to broadly advancing AI. Their release of FlashAttention v2 in summer stands as a significant milestone, garnering widespread adoption by leading AI research labs such as OpenAI, Anthropic, Meta, and Mistral. Additionally, their pioneering work in inference, utilizing novel techniques like Medusa and Flash-Decoding, has led to the development of the fastest inference stack for transformer models.
Regarding infrastructure, the startup boasts 20 exaflops across multiple data centers, highlights its readiness to handle the growing demands of AI applications.
The infrastructure, powered by NVIDIA GPUs and networking across partners like Crusoe Cloud and Vultr, is specifically designed for high-performance AI applications, offering cost-effective solutions for pre-training and inference workloads.
The platform is already being used by leading AI startups like Pika Labs, NexusFlow, and VoyageAI to develop next-generation generative models. Its inference API also allows one-click access to over 100 open source models for use in applications.
“Emergence invested early in the enterprise shift to cloud applications. We see a similar shift today where enterprises are rapidly investing in generative AI,” explains Joseph Floyd of Emergence Capital on his firm's investment. "Together AI is well positioned to be the platform of choice as enterprises look to control their proprietary IP while pushing their generative AI investments from prototype into production."
The founders of Together AI view generative models as a new operating system that will fundamentally reshape software development. By providing the tools for any developer or organization to build on open and custom AI, they aim to ensure access and interoperability within our AI future.
"Our mission is to create a way for any researcher or developer to participate in shaping our AI future," concludes CEO Prakash.
With renowned backers providing the resources to scale globally, Together AI’s open-focused approach looks set to expand options for enterprises seeking to leverage AI, unencumbered by single-vendor lock-in.