# NVIDIA Run:ai **Canonical:** https://apis.io/providers/runai/ **Website:** https://www.nvidia.com/en-us/software/run-ai/ **APIs profiled:** 62 NVIDIA Run:ai (formerly run.ai) is an AI operations and GPU orchestration platform for Kubernetes that pools, schedules, and governs GPU compute across clusters for training, fine-tuning, and inference workloads. It provides fractional GPU sharing, dynamic scheduling, quota and policy management, multi-tenant projects and departments, and workload lifecycle control across SaaS, self-hosted, and multi-tenant deployments. run.ai was founded in Israel, backed by Insight Partners and other investors, and acquired by NVIDIA in 2024; the product is now delivered as NVIDIA Run:ai. Its control-plane REST API exposes programmatic management of clusters, node pools, projects, departments, workloads (workspaces, trainings, inferences, distributed), assets, policies, permissions, service accounts, and audit logs, authenticated with bearer JWT access tokens obtained from client-credentials access keys. ## Kin Score — 47.8 / 100 (developing) Scored 2026-08-20 under rubric 0.12.0. Trend: flat (+0.0 from 47.8). | Facet | Score | |---|---| | Discoverability | 87.0 | | Contract Quality | 59.9 | | Governance | 16.7 | | Contract Governance | 16.7 | | Operational Transparency | 26.3 | | Developer Ergonomics | 66.1 | | Commercial Clarity | 27.6 | | Access Clarity | 27.6 | ## Agent readiness — 38.2 (agent-ready) | Dimension | Value | |---|---| | Spec Presence | yes | | Agentic Access | no | | Reversibility Documented | no | | MCP Server | no | | Auth Clarity | yes | | Idempotency | no | | Error Semantics | verified | | OpenAPI Examples | partial | | Rate Limit Signal | no | | Event Surface Described | no | | Agent Skills | derived | | Well Known Catalog | yes | | Consent Identity | no | | Agent Card | no | | Dry Run Mode | no | ## Access Self-serve signup — onboarding: self-serve, pricing: unknown, trial: no (confidence: medium). ## APIs (62) - **NVIDIA Run:ai Access Keys API** — Access Keys allow users to authenticate and interact programmatically with the NVIDIA Run:ai API. Each access key consists of a client ID and secret that can be used to obtain a... - **NVIDIA Run:ai Access rules API** — Access rules provide user authorization to system resources and entities. It is managed using Role-based access control (RBAC) which is a policy-neutral access control mechanism... - **NVIDIA Run:ai Administrator Command Line Interface API** — Information specific to the Administrator Command Line Interface. - **NVIDIA Run:ai AI Applications API** — AI Applications. - **NVIDIA Run:ai Applications API** — **DEPRECATED:** Applications have been renamed to Service Accounts. Please use the [Service Accounts](/latest/#tag/Service-Accounts) endpoints instead. Create and manage applica... - **NVIDIA Run:ai AuditLogs API** — The audit log provides audit trail information for user activity, changes to business objects and other important information. For more information, see [Audit log](https://run-... - **NVIDIA Run:ai Clusters API** — Use these endpoints to create, manage and delete NVIDIA Run:ai Kubernetes clusters. - **NVIDIA Run:ai Compute API** — A compute resource is a building block that represents compute resources such as GPUs, CPU cores, and CPU memory. The compute resources may consist of multiple physical resource... - **NVIDIA Run:ai ConfigMap API** — Use a ConfigMap as a data source location for data sets that are relevant to the workload being submitted. - **NVIDIA Run:ai Credentials API** — Credentials are used to unlock protected resources such as applications, containers, and other assets. For more information, see [Credentials](https://run-ai-docs.nvidia.com/saa... - **NVIDIA Run:ai Datasources API** — Data source assets. - **NVIDIA Run:ai Datavolumes API** — The Datavolumes API from NVIDIA Run:ai — 5 operation(s) for datavolumes. - **NVIDIA Run:ai Departments API** — Departments, in the hierarchy of resource allocation, are above Projects. A Department can contain multiple Projects, and has its own quotas. A Department's quota supersedes the... - **NVIDIA Run:ai Distributed API** — Distributed Training, is the ability to split the training of a model among multiple processors. It is often a necessity when multi-GPU training no longer applies; typically whe... - **NVIDIA Run:ai Distributed Inferences API** — Distributed inference enables running inference workloads across multiple pods, typically to scale model serving beyond a single container or node. This approach is useful when ... - **NVIDIA Run:ai Environment API** — An environment resource designates the container image, the image pull policy, working directory, security parameters, and others. It exposes all the necessary tools (open sourc... - **NVIDIA Run:ai Events API** — Workload events that occurred while the workload was running. Use to diagnose issue around workload scheduling. - **NVIDIA Run:ai Git API** — Use Git as a data source location for data sets that are relevant to the workload being submitted. - **NVIDIA Run:ai HostPath API** — Use a HostPath as a data source location for data sets that are relevant to the workload being submitted. - **NVIDIA Run:ai Idps API** — The Idps API from NVIDIA Run:ai — 3 operation(s) for idps. - **NVIDIA Run:ai Inferences API** — Inference workloads deploy trained models into a production environment to generate predictions from live data. These workloads are prioritized over Trainings and Workspaces dur... - **NVIDIA Run:ai Logo API** — Use to manage tenant logo files. - **NVIDIA Run:ai Me API** — "Me" returns the authenticated user's permissions within the system. It provides a comprehensive view of access rules (roles, subjects and scope) assigned to the current user. F... - **NVIDIA Run:ai Network Topologies API** — The Network Topologies API enables administrators to reflect the hierarchical network topology connectivity of nodes in a data center, such as racks, blocks, and other organizat... - **NVIDIA Run:ai NFS API** — Use NFS as a data source location for data sets that are relevant to the workload being submitted. - **NVIDIA Run:ai NodePools API** — Node pools assist in managing heterogeneous resources effectively. A node pool is a set of nodes grouped into a bucket of resources using a predefined (for example, GPU-Type) or... - **NVIDIA Run:ai Nodes API** — Nodes are worker machines in Kubernetes and may be either a virtual or a physical machine, depending on the cluster. Each Node is managed by the NVIDIA Run:ai control plane. For... - **NVIDIA Run:ai Notification State API** — Use to manage notification state. - **NVIDIA Run:ai Notification Types API** — Use to get notification types. - **NVIDIA Run:ai NotificationChannels API** — Notification Channels are the medium through which notifications are sent. - **NVIDIA Run:ai NVIDIA NIM API** — The NVIDIA NIM API provides endpoints to create and manage workloads that deploy NVIDIA Inference Microservices (NIM) through the NIM Operator. These workloads package optimized... - **NVIDIA Run:ai Org unit API** — Org unit. - **NVIDIA Run:ai Permissions API** — The Permissions API from NVIDIA Run:ai — 2 operation(s) for permissions. - **NVIDIA Run:ai Pods API** — Retrieve data about workload pods from your NVIDIA Run:ai platform. - **NVIDIA Run:ai Policy API** — Policies allow administrators to impose restrictions and set default values for researcher workloads. Restrictions and default values can be placed on CPUs, GPUs, and other reso... - **NVIDIA Run:ai Projects API** — Projects implement resource allocation policies and create segregation between different initiatives. It can represent a team, an individual, or an initiative that shares resour... - **NVIDIA Run:ai PVC API** — Use a PVC as a data source location for data sets that are relevant to the workload being submitted. - **NVIDIA Run:ai Registry API** — Use an images registry to enable the listting of repositories and tags that can be used as a data source location for data sets that are relevant to the workload being submitted. - **NVIDIA Run:ai Reports API** — The Reports API from NVIDIA Run:ai — 5 operation(s) for reports. - **NVIDIA Run:ai Researcher Command Line Interface API** — The Researcher Command Line Interface API from NVIDIA Run:ai — 9 operation(s) for researcher command line interface. - …and 22 more, listed in full on the page. ## MCP servers (1) - **NVIDIA Run:ai MCP Server** ## Security (2) - **Runai Authentication** — http · 1 scheme - **Runai Domain Security** — TLSv1.3 · HSTS · DNSSEC · DMARC ## Tags Company, Artificial Intelligence, GPU, Machine-Learning, Kubernetes, Orchestration, MLOps, Compute, Scheduling, Infrastructure --- Profiled by [API Evangelist](https://apievangelist.com) and published on [APIs.io](https://apis.io/providers/runai/). Scores are computed from the provider's own public artifacts under a published rubric.