Pricing and billing

Free models

The ways to call a model for free, their token limits, and what happens when an allowance runs out.

There are three ways to use a model without paying list price: the Free core allowance, a model's free promotion, and models that are free on your plan. Each has its own allowance. Your own provider keys are never throttled by any of them.

Three kinds of free

KindWhoAllowanceWhen it runs out
Free coreCard or payment verified organizationsGPT-6 Luna, DeepSeek V4.1 Flash and Clef: 10 requests and 250,000 input tokens per minute each; 1 million input plus output tokens per day, sharedPauses, or continues on credits if "Keep going on credits" is on
Free promotionSet per promotion, shown on the model pageHourly and daily token windowsStops with free_limit_reached until the window resets
Free on your planPro, Max and UltraThe plan's daily included allowanceBills at list price, or pauses

Daily windows reset at 00:00 UTC; hourly windows at the top of the hour.

At a glance
NeedsA verified card (the $1 hold) or any payment
Limits stated inTokens
Daily reset00:00 UTC
Free-only spelling<slug>:free

The :free spelling

A model whose free lane is a promotion can be called as <slug>:free. That spelling is free only: it is served while the allowance lasts and refused with free_limit_reached after, never charged.

You callAllowance leftAllowance spent
<slug>:freeFree429 free_limit_reached, never charged
<slug>Free429 free_limit_reached for a promotion; a Free core or plan-free model may continue on credits

With your key, GET /v1/models lists the :free spellings your organization can use.

{
  "error": {
    "type": "rate_limit_error",
    "code": "free_limit_reached",
    "message": "free_limit_reached: you used this model's free allowance of about 750,000 tokens this hour. It resets at 15:00 UTC. ..."
  }
}

The message names the exact window, its size in tokens and its reset. A window authored in money is converted to tokens at the model's list rates and marked "about".

Find free models

The free directory lists every place a model can be called for free, ours and other providers', with each provider's published limits. It is a directory, not the callable listing: entries marked serving: "external" are called at that provider with that provider's account.

GET/api/models
curl "https://api.experientiallabs.ai/api/models?free=true"
servingCall it
experientialThrough this gateway with your key
externalAt the provider's own base_url
CodeMeaning
free_limit_reachedA free window or allowance is spent
free_tier_requires_paymentThe free tier needs a card on file or a verified card
promo_byok_onlyThe promotion does not cover the full request and your balance cannot cover the rest

See Credits and billing for verification and Errors for every code.

Want no surprises? Call <slug>:free from scripts that must never spend credits.