product · May 22, 2026
CoreWeave Sandboxes Enters Public Preview with Serverless Access via Weights & Biases
Share the canonical public link.
CoreWeave announced CoreWeave Sandboxes in public preview on May 14, 2026, as an execution layer for reinforcement learning, agent tool use, and model evaluation. The offering runs isolated environments on customer CoreWeave Kubernetes Service clusters or as a serverless runtime through Weights & Biases using an existing W&B API key and Python client. Every sandbox operates in a fully isolated virtual environment by default with session management, storage integration, and monitoring captured in W&B run views; IBM Research and Mistral AI provided quotes on parallel sandbox usage at scale. Chen Goldberg, EVP Product and Engineering at CoreWeave, noted the layer closes the execution gap without custom infrastructure.