Together AI is a cloud platform built around open-source foundation models, providing the full development lifecycle: inference, fine-tuning, pre-training, and access to GPU clusters. It was founded in 2022 by Vipul Ved Prakash, Ce Zhang, Chris Re, and Percy Liang, a team combining startup founders with Stanford researchers. Prakash had earlier founded Topsy, a social-media search company acquired by Apple.
The company markets itself as “The AI Native Cloud” and offers serverless inference, batch processing, dedicated deployments, and GPU clusters ranging from instant self-serve instances to thousands of chips. Its advantage is framed through systems research: Together’s people are associated with work such as FlashAttention and custom inference kernels, which it cites in performance claims like faster inference and lower cost relative to naive serving. Customers referenced on its site include Cursor, Cohere, and ElevenLabs. The company has raised hundreds of millions of dollars at a multibillion-dollar valuation.
Why business readers should care: Together AI is one of several “neoclouds” competing to host and serve open-weight models more cheaply than the large hyperscalers. For teams that want to run open models without operating their own GPUs, these providers are the practical delivery channel, and their pricing and optimization directly affect the cost of building on open AI.