Credit, pricing & payment

Use the right AI. Pay for what you use.

Not every job needs the same model, the same reasoning level or the same amount of resources.

F68 AI lets users choose the AI that fits each task, and Credit is charged according to actual usage.

What is F68 Credit?

F68 Credit is the platform’s shared billing unit.

Add credit to your account and use the same balance across F68 AI models and agents.

One account. One Credit balance. Many AI capabilities.

Each completed task records the actual amount of resources used and deducts the corresponding Credit.

Top up Credit & pay in local currency — Details here →

The process:

Choose the top-up amount → check the Credit you will receive → pay → the credit is added to your account.

F68 aims to support payment in the local currency of each market.

Before confirming a transaction, users need to see clearly:

Top-up amount · Exchange rate applied · Taxes/fees by country · Credit received

Depending on the country and payment method, a transaction may incur tax, VAT/GST or other mandatory fees under local regulations. These must be shown clearly before the user confirms payment.

Final amount to pay = Credit value + Taxes/fees applicable in the user’s country.

The exchange rate, taxes and payment method are determined from actual data at the time of the transaction.

Credit is deducted according to actual usage

The cost of a task can depend on:

• model;
• reasoning level;
• amount of input data;
• length of the result;
• context;
• tools used;
• the actual resources needed to complete the task.

Light tasks use less Credit. Complex tasks that use more resources use more Credit.

Users can actively balance:

quality · speed · depth · cost.

Track usage — Details here →

F68 AI lets users check the usage of each task and their account history.

Users can see which model was used, how many resources were consumed, how much Credit was deducted, and total usage over time.

Beyond knowing the remaining balance, users also know where their Credit is being spent.

Context also affects cost

The longer a conversation, the more context the AI has to process in each task.

F68 AI lets users track context by:

System Prompt · Context · Tool · History

When context becomes large, users can compress the conversation, clear context, clear tool data or branch the chat.

See details in 09. Managing conversation context.

Caching can also affect cost

Caching is a mechanism of the model/API provider that reuses input or context that was already processed earlier when a new request has content similar enough and meets the model’s conditions.

A cache hit may be charged by the provider at a lower input rate than a cache miss.

F68 AI does not build its own content cache store for this feature.

F68 AI shows and uses the caching mechanism of the corresponding model/API, according to the configuration the provider supports.

Caching is most valuable when the same part of the context is reused many times, especially with:

• AI Agents with long System Prompts;
• large context;
• multi-turn conversations;
• workflows that reuse the same documents.

See: Using caching effectively →

The strongest model isn’t always the best choice

F68’s principle:

Use just enough intelligence for the job.

Simple tasks can use a lighter model. Research, analysis or work where mistakes are costly may need a stronger model and deeper reasoning.

The job decides how strong the AI needs to be. The user decides how many resources to use.

Purchased credit is non-refundable

Once purchased credit has been added to your account, it cannot be refunded in cash.

If you experience a transaction issue or the credited amount is incorrect, contact support to investigate and reconcile the transaction.

Reconciling a transaction does not mean refunding Credit that was topped up.

Model pricing

Rates per 1 million tokens: