Credits

Every request spends one of three credits. Which one depends on the model you call.

Daily free credits

A :free model costs 1 daily free credit per request, and never touches your premium credits.

Basic (free) plan: every free model has its own daily limit, and the limit varies by model.
Standard and up: unlimited, with no per-model cap.

Ordinary :free variants are capped at 5,000 input tokens on every plan. A listed free variant with a context suffix uses that variant's context budget.

Daily premium credits

These pay for premium, :metered, :full-context, and :metered:full-context requests. Your plan sets the daily allowance and it resets at 00:00 GMT+7, with higher plans getting more.

Named context variants such as :32k-context also use premium credits at a flat price per request, unless the ID includes :free or :metered. What a single request costs depends on its variant and its length. The cost calculator works it out, and model naming gives the rule behind it.

Permanent premium credits

A standing balance for any model that spends premium credits. It spends itself once the day's allowance runs out, so your requests keep working. It never resets and only goes down when used.

Topping up

Buy permanent premium credits on the Billing page of your dashboard, under Top Up Permanent Credits:

$1 buys 5,000 credits.
Each top-up is a whole-dollar amount from $3 to $500.
Pay by card through Stripe, or in crypto (USDT or USDC) with a 5% discount.

Top-ups work on every plan, Basic included.

Pay as you go

A :payg model is billed per token instead of per request. Input tokens, output tokens, and cached input tokens each have their own price per 1M tokens, shown on the model's page under Models. $1 of that price is 5,000 credits.

Paid plans: your daily premium credits pay first, then your permanent premium credits.
Basic (free) plan: paid from permanent premium credits only. A request needs at least 0.1 credits to start. The cost is only known once the answer is complete, so the balance can dip below zero for a moment; your next top-up settles it.
Every plan can use :payg models, and your chat memory settings never trim them.

Where to check them

All three balances are on the Usage Overview page of your dashboard.