NewHow we catch a seller listing the wrong model, and refund you if one does.

Authentication

API keys

Inference requests authenticate with a MartLLM API key. Keys start with ull_ followed by a random suffix. Pass it as a bearer token:

Authorization: Bearer ull_your_key_here

The Anthropic surface also accepts the key in the x-api-key header.

Your first key is made for you on your first visit to the dashboard and shown there with a first call you can paste. Create, edit, list, and delete keys under Dashboard → API Keys. The full secret is displayed once at creation — store it immediately; MartLLM keeps only a hash and can never show it again. Deleting a key (Dashboard → API keys → Delete) takes effect at once: apps using it stop working, and its past usage stays in your totals.

Per-key controls

Each key can carry its own guardrails, set when you create it or later with Edit on Dashboard → API keys. A change applies to the key's next request.

  • Spend limit — a cap in USD over a daily, monthly, or lifetime window. Requests that would exceed it are rejected with 402 key_spend_cap_exceeded. Empty means no limit.
  • Daily limit — daily_limit_usd, on top of the spend limit. A request must fit under both, and the daily limit resets at midnight UTC. A request over either limit is refused with 402 key_spend_cap_exceeded and nothing is charged. Set it on the Keys page or with PATCH /api/v1/keys/{id}.
  • Rate limit — a per-minute request ceiling (1 to 10,000). Empty means the standard per-key limit (300 a minute).
  • Do not save my requests and answers — off by default: MartLLM keeps each request and its answer for 90 days to fix failed requests and answer support questions. Turn it on and calls made with the key are not saved. We keep only the record of each call (time, model, token counts, cost, the address it came from, the providers we tried and any error they returned), not what you sent or what came back, so we cannot look into a wrong answer for you. Video and music jobs still keep their request, so a job a provider drops can be started again. The provider that serves a request receives it either way. Set it on the Keys page, or with no_store (true/false) on POST /api/v1/keys and PATCH /api/v1/keys/{id}; GET /api/v1/keys returns it.
  • BYOK provider — pin the key to route through your own provider credential (see below).

Low-balance email

On Dashboard → Credits you can ask to be emailed when your balance falls below an amount you choose. It is off by default, sends one email each time the balance drops below that amount, and goes to the account's owners. If money set aside for an image, video, music or speech request is still held 15 minutes after the job ended, the account owners get an email listing those jobs and our team is told. That money is still yours and will be released.

BYOK — bring your own key

Add a provider credential under Dashboard → BYOK. Requests routed through a BYOK provider use your credential directly; no credits are reserved and the per-token cost is zero on your MartLLM balance. Stored credentials are encrypted and only ever surfaced as a masked last-4.

Dashboard sessions

You sign in to the web dashboard with Google (one click: the "Continue as" prompt on martllm.com, or the Google button), or with your email: we send a 6-digit code (and a sign-in link) that expires in 15 minutes. There is no password. Signing in mints a short-lived session, stored in an httpOnly cookie. This session is separate from your ull_ inference keys — dashboard endpoints under /api/v1/* use the cookie; the gateway (/v1/*) uses the API key.

Errors

Auth failures return an OpenAI-style error envelope:

{ "error": { "code": "invalid_api_key", "message": "…", "type": "authentication_error" } }

See Errors & limits for the full list.