Mafdet AI Help Center中文

Subscription plans

Your subscription plan is what you buy. It sets your Playground credit, your rate limits, how many API keys you can hold, and which models your API keys may call.

A separate concept — the model access tier — is a property of each model, not of you. Plans grant access to tiers. The two are easy to confuse, so this page keeps them apart.

Plans

PlanPricePlayground credit / periodRate limit (tpm)Max API keys
FreeContact us$0.5012,0003
EnterpriseContact us$500.001,000,000Unlimited
Starter$10.00 / month$9.5060,0003
Pro$30.00 / month$28.50300,00025

Plans are self-service: buy, upgrade or cancel from the Subscription page in the console. You do not need to contact an administrator.

What each plan can call

Playground and API keys have separate model access, and this catches people out:

PlanModels in PlaygroundModels your API keys can call
FreeFree model onlyFree model only
StarterAll public modelsFree model only
ProAll public modelsFree + standard + advanced models
EnterpriseEverythingEverything

So Starter is a Playground plan: you get the full model line-up in the browser, paid from your $9.50 Playground credit, but an API key issued on Starter can only call the free trial model. Calling paid models from the API starts at Pro.

The free trial model is gemini-3-flash-lite (billed at $0). Everything else is a paid model.

Signup bonus: while promotional bonus remains in your wallet, paid models are temporarily unlocked so you can try them. When the bonus is used up the restriction returns — this is a trial, not a plan entitlement.

Model access tiers (background)

Each model declares the tier required to call it: FREE, PAID, TRUSTED, ENTERPRISE, plus an INTERNAL_BETA marker that stays hidden from ordinary users. Your plan grants a set of these tiers, separately for Playground and for API keys — the table above is that mapping in plain language. Each model's tier is shown on the Models overview.

Plan limits you will actually notice

  • Rate limit (tpm) — tokens per minute. A single large request can exceed it on its own: a video sent to the embeddings API is base64 (roughly 929 tokens per KB), so video embedding effectively needs Pro or above. See the 429 entry in Errors.
  • Per-request cost cap — a request that would cost more than the cap is refused before it runs.
  • Max API keys — how many active keys you can hold at once.