Skip to main content

Free tier, no signup

The VLM Run Gateway serves anonymous callers on a small free quota, keyed by client IP. An API key raises the quota and attributes usage to your account. See Rate Limits for the per-tier limits.
The VLM Run Gateway accepts a standard Authorization: Bearer <token> header on every endpoint. The header is optional: a call with no header is anonymous, so OpenAI(base_url=...) works without an API key. Bearer vlmrun is the explicit anonymous form for clients that need a non-empty key value.

​
Tiers

Authenticating raises your quota and changes attribution: anonymous requests are capped per IP (so they can be shared with other users behind the same NAT or proxy), while authenticated requests get the higher limit above, capped per account.

​
Tier resolution

The Gateway resolves your tier from the Authorization header in this order:
  1. No header → anonymous, keyed by client IP. A deployment can set VLMRT_REQUIRE_AUTH_HEADER=1 to reject a missing header with 401, {"detail": "Missing API Key"}. The hosted Gateway does not set it.
  2. Bearer vlmrun, or an empty bearer value → anonymous. vlmrun is the documented sentinel, and the preferred form, because SDKs such as the OpenAI Python client reject an empty api_key before the request leaves the process.
  3. Valid VLM Run API key → authenticated, as the owning user and organization.
  4. Any other non-empty bearer token → 403 (invalid API key). The Gateway does not fall back to anonymous access for unrecognized tokens.

​
Credits

An authenticated call is charged to the caller’s organization. When that organization has no credit left, the Gateway answers 402:
The routed models carry the paid access tier, so they need an organization with an active subscription or a funded balance. Every GPU-served model is public and reachable anonymously.
Every endpoint documented under API Reference accepts anonymous traffic today: chat completions, models, embeddings, audio transcriptions, and health. Authenticate with an API key for anything you plan to run in production.

​
Get an API key

  1. Sign up at app.vlm.run.
  2. Copy your API key from Settings → API keys.
  3. Set it as an environment variable:

​
Examples

An invalid (non-empty, unrecognized) bearer token returns 403 rather than falling back to anonymous access. See Error Codes for the response bodies.

Rate Limits

Per-tier quotas and how to request a higher limit.

Error Codes

402, 403, and 429 response bodies.