Together, a startup specializing in open source generative AI and AI model development infrastructure, has successfully closed a $102.5 million Series A funding round led by Kleiner Perkins. Notable participants in the round include Nvidia and Emergence Capital. The significant funding injection is expected to support the expansion of Together’s cloud platform, which enables developers to build on open and custom models. This latest round of funding is over five times larger than the company’s previous raise.
Key Takeaway
Together, a startup focused on open source generative AI, has secured a
02.5 million investment in a Series A funding round. The funding will be used to expand Together’s cloud platform for developers to build on open and custom models. CEO Vipul Ved Prakash envisions generative AI as a platform technology with long-lasting societal impact. Together’s cloud platform offers scalable compute at lower prices compared to major vendors, and the company also provides custom model design services. Additionally, Together actively engages in AI research, releasing open source generative models and fostering innovation in the field.
A Vision for Generative AI
In a blog post on Together’s website, co-founder and CEO Vipul Ved Prakash emphasized the growing interest among startups and enterprises in developing a generative AI strategy that is not tied to a single vendor. Prakash believes that open source AI provides a strong foundation for these applications, with increasingly powerful generative models being released on a regular basis. He sees generative AI as a platform technology, akin to a new operating system for applications, and foresees its significant and far-reaching impact on society. Prakash emphasizes the importance of choice and options in the AI ecosystem, envisioning a future consisting of both proprietary and open models.
The Team Behind Together
The inception of Together took place in June 2022, founded by Vipul Ved Prakash alongside Ce Zhang, Chris Re, and Percy Liang. Prakash, known for previously founding social media search platform Topsy, which was later acquired by Apple, brings a wealth of experience to his role as CEO. Together’s mission is to develop open-source models and services that facilitate the integration of AI into organizations’ applications.
Cloud Platform and Services
To support their mission, Together has built a cloud platform that enables the running, training, and fine-tuning of open source models. The co-founders claim that this platform offers scalable compute at lower prices compared to dominant vendors such as Google Cloud, AWS, and Azure. Together currently operates a cloud infrastructure across datacenters in the U.S. and EU, including servers operated by partners Crusoe Cloud and Vultr. This infrastructure delivers approximately 20 exaflops of compute in total.
The company counts Pika Labs, NexusFlow, Voyage AI, and Cartesia among its customers, who not only utilize Together’s cloud platform but also leverage their model-serving APIs. Vipul Ved Prakash highlighted the advantages of their custom infrastructure, which offers better economics for pre-training and inference workloads. Together’s AI platform allows developers to integrate leading open source models or create their own models through pre-training or fine-tuning. Customers maintain ownership of their investment in AI and have the freedom to run their models on any platform.
Complementary Services and AI Research
In addition to their cloud platform, Together provides a consulting service called “Custom Models.” This service allows customers to bring their own data to the Together cloud platform, where the company’s team works on designing, building, and testing a model tailored to the customer’s specific needs.
Together is also heavily invested in AI research. One of their notable projects, RedPajama, aims to cultivate a set of open source generative AI models, including “chat” models akin to OpenAI’s ChatGPT. The company has also released a fine-tuned version of Meta’s Llama 2 text-generating model, as well as GPT-JT, a fork of the open source text-generating model GPT-J-6B from EleutherAI’s open research group. Another endeavor, OpenChatKit, is an attempt at developing a ChatGPT-equivalent model.
