xAI (Grok)

Full account visibility for xAI. Surfaces key metadata, credit breakdown, rate limits, and the list of models the key is allowed to call.

At a glance

• Provider ID — xai
• Detection — XAI_API_KEY environment variable
• Auth — API key
• Type — API platform (full billing data)
• Tracks:
  • Key info: name, team
  • Credits: remaining, spent, granted (USD)
  • RPM and TPM
  • Allowed models

Setup

Auto-detection

Set XAI_API_KEY.

Manual configuration

{
  "accounts": [
    {
      "id": "xai",
      "provider": "xai",
      "api_key_env": "XAI_API_KEY",
      "base_url": "https://api.x.ai/v1"
    }
  ]
}

Data sources & how each metric is computed

Each poll (default every 30 seconds in daemon mode) makes two calls under https://api.x.ai/v1. All requests use Authorization: Bearer $XAI_API_KEY.

Call	Endpoint	What it provides
1	GET /api-key	Key metadata, credit breakdown, allowed models
2	GET /models	Rate-limit headers

Key metadata

• Source: top-level fields of the /api-key JSON: name, team_id, id.
• Transform: stored under Raw["api_key_name"], Raw["team_id"]. Used for the tile heading.

credits — combined credit metric

• Source: /api-key fields remaining_balance, spent_balance, total_granted.
• Transform: copied as Remaining, Used, Limit of a single metric. Currency is fixed at USD.
• The status message becomes $X.XX remaining (formatted from Remaining).

Allowed models

• Source: allowed_models array on /api-key.
• Transform: stored as Raw["allowed_models"]. The detail view lists them; calls to other models fail at xAI's edge.

rpm / tpm — rate limits

• Source: response headers on GET /v1/models
  • x-ratelimit-limit-requests, x-ratelimit-remaining-requests, x-ratelimit-reset-requests
  • x-ratelimit-limit-tokens, x-ratelimit-remaining-tokens, x-ratelimit-reset-tokens
• Transform: parsed verbatim.

Auth status

• Source: HTTP status code. 401/403 → auth; 429 → limited; otherwise ok.

What's NOT tracked

• Promo vs paid split. total_granted lumps promotional and paid credits together. The API does not break them apart.
• Per-model spend. The credit endpoint returns aggregate dollars only.

How fresh is the data?

• Polled every 30 s by default. The credit endpoint reflects xAI's near-real-time accounting.

API endpoints used

• GET /v1/api-key
• GET /v1/models

Caveats

• Granted credits include both promotional and paid; the API does not split them further.
• Allowed models reflect the key's scope, not the team's full catalog.
• Currency is USD.

Troubleshooting

• Empty allowed models — the key has no model permissions; create a new key with model access in the xAI console.
• Spend higher than expected — xAI charges for both successful and certain failed requests; check the console for itemized billing.