What BentoML does
BentoML is an open-source and enterprise inference platform used by ML teams to package, deploy, and operate model inference in production. The company’s open-source core (the BentoML Python framework) focuses on turning Python-defined “Services” into deployable inference workloads, while its commercial platform (BentoCloud and the broader “Bento Inference Platform” marketed on bentoml.com) targets production needs such as deployment automation, observability, scaling behavior, and deployment control across cloud and on-prem environments.
BentoML’s technical approach centers on creating a standardized “Bento” artifact that bundles application code, dependencies, and model artifacts so teams can reproduce deployments across environments. The BentoML documentation describes “Bentos” as the standardized packaging format for AI/ML services and provides workflows for building, deploying, and containerizing those artifacts.
On the enterprise side, BentoCloud is positioned as an inference management platform and compute orchestration engine built on top of BentoML’s open-source serving framework. BentoCloud’s documentation and product messaging emphasize cloud deployment workflows, support for GPUs, and operating inference systems across development, testing, deployment, monitoring, and CI/CD.
As of February 2026, BentoML’s strategic position shifted materially: BentoML was acquired by (joined) Modular. BentoML and Modular both state that BentoML remains Apache 2.0 and that existing customers are supported without disruption while the teams integrate deeper into Modular’s stack.