Guides

Billing

Prepaid balance

The API is pay as you go, in reais (BRL). Each request is debited from the account balance by the tokens used, at the price of each model in Models. New accounts get R$1.00 in credits.

The Chat (R$39.90/month) and Chat Pro (R$69.90/month) plans cover the chat at chat.offmem.app. API requests always use the balance.

How cost is calculated

Prices are per 1 million tokens. Cached input tokens are charged at the cached input price. The amount is rounded up to the millionth of a real.

cost = (input − cached) × input price
     + cached × cached input price
     + output × output price

GLM 5.3 Flash, 2,000 input tokens and 500 output tokens:
2,000 × R$1.25 / 1M + 500 × R$5.15 / 1M = R$0.005075

Reserve

Before calling the model, offmem reserves an estimate of the request cost. After the response, the reserve is replaced by the real cost reported by the model.

• Input: 1 token for every 3 characters of the request.
• Images: 1,500 tokens each.
• Output: max_tokens (or max_completion_tokens), or 4,096 tokens when not sent, up to 32,768.

If the available balance is lower than the reserve, the request returns 402 sem_saldo and nothing is charged. A lower max_tokens lowers the reserve.

What is not billed

• Requests the model answers with an HTTP error, or that never reach the model.
• GET /v1/models and POST /v1/messages/count_tokens.

A stream closed before the end is billed by estimate: the input plus what was already delivered.

Top-up

Referrals

• Whoever refers gets 15% of each payment of the referred account, in credits, for 12 months after sign-up.
• The referred account gets 25% more credits on its first payment, up to R$50.00.

Referral link in Console › Referrals.

Spend control

• Daily and monthly cap per key, in reais: Key options.
• Each request, with model, tokens and cost, in Console › Activity, with CSV export.