Baseten

Baseten provides infrastructure for running AI models in production. Engineering teams deploy open-source, custom or fine-tuned models on dedicated GPU capacity, and Baseten handles autoscaling, cold starts, multi-cloud capacity and high availability, either in its own cloud or in the customer's. Its inference stack includes optimized model runtimes, custom kernels and caching to reduce latency and cost. Developers can also call popular open models through pay-per-token Model APIs, and train models with reinforcement learning before deploying them on the same stack. The company works with AI product companies, enterprises and model labs that need fast, reliable model serving at scale.