Skip to main content

Mistral AI

Full billing visibility for Mistral AI. Surfaces the monthly budget, credit balance, monthly spend, token totals, and rate limits — all in EUR.

At a glance​

  • Provider ID — mistral
  • Detection — MISTRAL_API_KEY environment variable
  • Auth — API key
  • Type — API platform (full billing data)
  • Tracks:
    • Plan
    • Monthly budget (EUR)
    • Credit balance (EUR)
    • Monthly spend
    • Monthly tokens (input and output)
    • RPM and TPM

Setup​

Auto-detection​

Set MISTRAL_API_KEY.

Manual configuration​

{
  "accounts": [
    {
      "id": "mistral",
      "provider": "mistral",
      "api_key_env": "MISTRAL_API_KEY"
    }
  ]
}

Data sources & how each metric is computed​

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

CallEndpointWhat it provides
1GET /billing/subscriptionPlan name, monthly budget cap, credit balance
2GET /billing/usage?start_date=YYYY-MM-01&end_date=<today>Daily spend & tokens for the current month
3GET /modelsRate-limit headers (RPM, TPM)

monthly_budget — plan cap​

  • Source: monthly_budget field of /billing/subscription.
  • Transform: copied verbatim into Limit. Currency: EUR.

credit_balance — remaining credits​

  • Source: credit_balance field of /billing/subscription.
  • Transform: copied verbatim into Remaining. Currency: EUR.

monthly_spend — month-to-date cost​

  • Source: total_cost from /billing/usage for start_date = first of the current UTC month, end_date = today.
  • Transform: stored as Used. If monthly_budget is known, Limit is set to it and Remaining = Limit - Used. Currency: EUR.

monthly_input_tokens / monthly_output_tokens​

  • Source: sum of input_tokens / output_tokens across every entry in data[] returned by /billing/usage for the current month.
  • Transform: simple row-by-row sum. Stored as raw token counts.

rpm / tpm — rate limits​

  • Source: response headers on GET /v1/models. Three header groups are read:
    • Primary rpm — ratelimit-limit, ratelimit-remaining, ratelimit-reset (no x- prefix).
    • Primary tpm — x-ratelimit-limit-tokens, x-ratelimit-remaining-tokens, x-ratelimit-reset-tokens.
    • rpm_alt — x-ratelimit-limit-requests, x-ratelimit-remaining-requests, x-ratelimit-reset-requests. Mistral occasionally returns this alongside the primary headers; OpenUsage exposes it as a separate metric so both are visible.
  • Transform: parsed verbatim into the corresponding metrics.

Auth status​

  • Source: HTTP status code on any of the three endpoints. 401/403 → auth; 429 → limited; otherwise ok.

What's NOT tracked​

  • Per-model breakdown. /billing/usage returns daily aggregates; the provider sums them month-to-date and does not split by model.

How fresh is the data?​

  • Polled every 30 s by default. The /billing/usage totals are themselves aggregates Mistral updates on its own cadence — typically a few minutes behind real time.

API endpoints used​

  • GET /v1/billing/subscription
  • GET /v1/billing/usage?start_date=…&end_date=…
  • GET /v1/models

Caveats​

warning

Mistral bills in EUR. Mixing it with USD-billed providers in a single dashboard requires you to convert manually.

  • The billing period is the calendar month; numbers reset at midnight UTC on the 1st.
  • Rate-limit headers come from /v1/models.

Troubleshooting​

  • No spend data — verify the API key has billing scope; check Mistral's console.
  • Currency confusion — Mistral always reports EUR; OpenUsage displays whatever the API returns.

Why doesn't monthly spend match the Mistral console exactly?​

The dashboard sums data[].total_cost from /v1/billing/usage for [first-of-month, today]. Mistral's console can include same-day usage that hasn't aggregated yet, or apply different rounding. Refresh after the next aggregation pass.