Baseten

Baseten provides infrastructure for developers and enterprises to train, deploy, and scale AI models in production. Its Dedicated Inference serves custom and fine-tuned models, while Model APIs provide pre-optimized models for testing and production use. Training runs on the same infrastructure stack as inference. Teams can deploy in Baseten’s managed cloud, their own VPCs, or a hybrid configuration, supporting applications such as image generation, transcription, and voice agents. Customers include Cursor, Notion, Abridge, and Harvey.

Founded in 2019 to address the difficulty of deploying and scaling machine learning systems, Baseten’s founders are CEO Tuhin Srivastava, CTO Amir Haghighat, Chief Scientist Phil Howes, and Pankaj Gupta. Its engineering contributions include Truss, an open-source Python package for packaging, serving, and deploying models, and Chains, a Python framework for workflows combining models and business logic. Chains gives each component its own hardware and autoscaling, allowing developers to separate CPU processing from GPU inference while retaining local testing and typed interfaces.

In June 2026, Baseten raised a $1.5 billion Series F at a $13 billion valuation. Its annualized revenue run rate reached an estimated $600 million as of March 2026. Baseten charges for usage, through API consumption or minutes and hours of GPU capacity.