For AI agents: a documentation index is available at the root level at /llms.txt and /llms-full.txt. Append /llms.txt to any URL for a page-level index, or .md for the markdown version of any page.
  • Getting Started
    • Overview
    • Install on Kubernetes
    • Install on Slurm
  • Architecture
    • Architecture
    • Config and API
    • Modeling and API Simulation
  • Providers
    • AWS
    • GCP
    • OCI
    • Nebius
    • Nscale
    • InfiniBand
    • NetQ
    • DRA
    • Test
  • Engines
    • SLURM
    • Kubernetes
    • Slinky
  • Reference
    • Node Labels and Annotations
NVIDIANVIDIA
Developer-friendly docs for your API
Privacy Policy | Your Privacy Choices | Terms of Service | Accessibility | Corporate Policies | Product Security | Contact

Copyright © 2026, NVIDIA Corporation.

LogoLogoTopograph
On this page
  • Using credentials:
  • Credentials via File
  • Credentials via API Request Payload
  • Using authorization token
Providers

Nebius Topology Provider

||View as Markdown|
Previous

OCI

Next

Nscale

The Nebius topology provider uses the Nebius AI Cloud SDK for Go. The Services().Compute().V1().Instance().List() method returns a list of compute instances for a specified project. Each instance may include a Status.InfinibandTopologyPath field, which is an array of three network IDs. If present, these IDs describe the path through the three-tier network, from the root switch down to the leaf switch.

To use the API, you must provide authorization. There are two ways to do this: using account credentials or an authorization token.

Using credentials:

Nebius credentials consist of the following fields:

  • projectId
  • serviceAccountId
  • publicKeyId
  • privateKey

The projectId field is optional. If it is omitted, the project ID is retrieved from /mnt/cloud-metadata/parent-id.

You can provide credentials either in the Topology configuration file or directly in the topology request payload.

Credentials via File

Store your credentials in a YAML file:

1projectId: <PROJECT-ID>
2serviceAccountId: <SERVICE-ACCOUNT-ID>
3publicKeyId: <PUBLIC-KEY-ID>
4privateKey: <PRIVATE-KEY>

Then reference this file in your Topograph config:

1http:
2 port: 49021
3 ssl: false
4
5provider: nebius
6engine: slurm
7
8credentialsPath: /path/to/credentials.yaml

Credentials via API Request Payload

Pass credentials directly in the topology request payload:

1\{
2 "provider": \{
3 "creds": \{
4 "projectId": "PROJECT-ID",
5 "serviceAccountId": "<SERVICE-ACCOUNT-ID>",
6 "publicKeyId": "<PUBLIC-KEY-ID>",
7 "privateKey": "<PRIVATE-KEY>"
8 \}
9 \}
10\}

Using authorization token

You can provide an authorization token in one of two ways:

  • Via the environment variable IAM_TOKEN
  • By placing the token in the file /mnt/cloud-metadata/token