Baseten

Baseten is an AI inference infrastructure company that enables developers to deploy and scale machine learning models with minimal operational overhead. Founded in 2019, the company provides a serverless GPU platform optimized for running AI models in production.

Baseten's platform supports popular model serving frameworks including vLLM, TensorRT-LLM, and Triton, with automatic scaling, GPU optimization, and built-in monitoring. The company specializes in making it easy to deploy open-source models (Llama, Mistral, Stable Diffusion) as production-ready API endpoints with sub-second cold starts and efficient GPU utilization. Baseten's Truss framework is an open-source model packaging standard that simplifies the path from model development to production deployment.

Baseten has raised rapidly escalating venture funding as the inference market has heated up — a $300 million Series E in early 2026, followed within months by a $1.5 billion Series F led by Altimeter Capital, Conviction, and Spark Capital that confirmed inference as one of AI's most contested infrastructure layers. By mid-2026 the platform handled more than 1 billion inference calls a day, with revenue up roughly 20 times year over year. The company serves customers ranging from AI startups to enterprises that need reliable, low-latency model inference without managing GPU infrastructure directly, and competes in the growing model inference market alongside Together AI, Replicate, Fireworks, and cloud-provider offerings.