Together AI

Together AI is a cloud platform for open and open-weight models, providing inference, fine-tuning, and training infrastructure optimized for cost and speed. It serves hundreds of models through a single API, contributes to open-source research (including efficient inference and long-context techniques), and positions itself as the neutral home for teams building on open models rather than closed frontier APIs.

Founded in 2022, Together AI set out to make open models practical to deploy at scale, combining research on inference optimization with a GPU cloud tuned for serving. Its team has published widely used open-source work on efficient attention and model training.

The platform offers a broad catalog of open-weight models (Llama, Qwen, DeepSeek, Mixtral, and many more) via an OpenAI-compatible API, plus dedicated endpoints, fine-tuning, and full training clusters. Its pitch is price-performance and control: run open models cheaply, customize them, and avoid lock-in to a single closed provider.

Backed by prominent venture investors at a multi-billion-dollar valuation, Together AI competes with Fireworks, Groq, Baseten, and the hyperscalers' model-serving offerings, and benefits directly from the rising quality of open-weight models from Meta, Alibaba, and DeepSeek.