OpenLease is an open-source orchestration tool designed to make rented GPUs programmable for deploying and managing open-source large language models (LLMs). It addresses the challenge of provisioning, operating, and safely tearing down GPU infrastructure for ephemeral, cost-sensitive workloads, such as those encountered in batch processing or AI inference tasks. OpenLease provides a lifecycle management system that automates the deployment, observation, recovery, proxying, metering, and teardown of GPU-backed model endpoints.
The platform supports deploying LLMs on GPUs, reconciling deployments to a desired state through a state machine model, and serving inference via an OpenAI-compatible API. It includes features such as a reconcile loop for crash recovery, cost-safety mechanisms to prevent idle billing, and an orphan sweep to clean up unused resources. Users can interact with OpenLease through three primary interfaces: a command-line interface (CLI), a REST API (with OpenAI proxy endpoints and bearer-token authentication), and an MCP server that enables agent-driven operations. The CLI provides commands for deploying models, checking deployment status, stopping deployments, monitoring costs, and managing health and availability. The REST API mirrors these lifecycle operations and exposes an OpenAI-compatible endpoint at /v1/.
The tool is built for scenarios where rapid, temporary access to GPU resources is required, such as video indexing or bursty AI workloads. It is validated on RunPod hardware, with RunPod serving as the initial GPU provider, though the architecture is provider-neutral and can support any vLLM-servable model. OpenLease offers features like model catalog browsing, real-time GPU availability checks, model caching for faster redeployments, and detailed status reporting, including cost projections and deployment health.
0 open-source license. The project is developed by Canonical, an independent AI consultancy. Its design emphasizes safe, agent-operable GPU control, making it suitable for AI practitioners and organizations managing dynamic, high-cost GPU workloads.
In the Infrastructure & Backend space, open-lease takes a focused approach. It focuses on automating the provisioning of GPU infrastructure and deployment of open-source LLMs for developers. It is built as an open-source project for machine learning engineers. open-lease is open source under the Apache-2.0 license. It runs on the web, the command line, and API, and it can be self-hosted.
mfbaig35r builds and maintains open-lease, and the product first shipped in 2026. Development happens publicly on GitHub with 38 commits in the last 90 days. PulseGate's similarity index finds few close equivalents — open-lease occupies a relatively distinct niche. Key capabilities include GPU provisioning, LLM deployment, and OpenAI-compatible API.
Latest indexed changes and source events
Other apps tracked under the same category.
