API keys

RemoteGPU API keys are used for two main purposes:

• calling inference APIs with x-api-key
• refreshing Kubernetes exec credentials for kubectl

Console account management APIs do not use API keys. Those APIs use your signed-in browser session instead.

Where to manage keys

Open Settings / API Keys in the console to:

• create keys
• rename keys
• change expiry
• change scopes
• revoke keys

The raw key value is returned only when the key is created. After that, the API and console only expose its prefix.

Scope shapes

There are three supported scope modes.

All access

• all

{
  "all": true
}

Use this when one key should work for both inference and Kubernetes flows.

Inference only

{
  "inference": true
}

Use this for image generation clients that should not be able to refresh Kubernetes credentials.

Kubernetes exec only

Namespace-scoped:

{
  "kubernetes_exec": {
    "namespace_ids": [
      "YOUR_NAMESPACE_UUID"
    ]
  }
}

All namespaces owned by the current account:

{
  "kubernetes_exec": {
    "all_namespaces": true
  }
}

Use this for kubectl access without granting inference APIs.

Invalid scope combinations

A mixed payload with explicit inference access and namespace-scoped Kubernetes access is not valid. Use {"all": true} when you want both.

These payloads are rejected:

• {}
  At least one explicit scope is required
• {"inference": true, "kubernetes_exec": {...}}
  Explicit mixed scopes are not supported
• {"all": true, "inference": true}
all cannot be combined with other scope fields
• {"kubernetes_exec": {"namespace_ids": []}}
  At least one namespace ID is required unless all_namespaces is true

Namespace ownership rules

kubernetes_exec.namespace_ids must reference namespaces owned by the current account. If you include a namespace from another account, creation or update is rejected with 400.

Header format

APIs expect the key in the x-api-key header:

curl -H "x-api-key: YOUR_API_KEY" \
  "https://api.remotegpu.ai/v1/inference/runtime/models"

Example inference request:

curl -X POST "https://api.remotegpu.ai/v1/inference/image" \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_API_KEY" \
  -d '{
    "model": "black-forest-labs/FLUX.1-dev",
    "prompt": "a cinematic portrait of an astronaut cat"
  }'

Kubernetes kubeconfig refresh also uses x-api-key, but the generated kubeconfig exec plugin normally handles that for you.

Common use cases

• Inference-only client integration
• Namespace-scoped kubectl access
• Broad admin automation with all scopes

Lifecycle rules

• Key names are limited to 64 characters
• Expiration must be in the future
• Revoked keys cannot be modified
• Expired keys cannot be modified
• Creating a key can fail with 409 if your account has reached its active key limit

Common responses

• 401 Missing access token when managing keys from account APIs
• 401 Missing, invalid, revoked, or expired API key when calling key-protected APIs
• 400 Invalid scope payload, invalid namespace ownership, or past expiration
• 404 API key not found
• 409 Revoked or expired keys cannot be modified

Related APIs

• GET /v1/account/api-keys
• POST /v1/account/api-keys
• PATCH /v1/account/api-keys/{key_id}
• POST /v1/account/api-keys/{key_id}/revoke