Credits are the usage balance for model services in PlugClaw. When a message is sent, content is generated, or another AI-powered task is performed, the model processes a certain number of Tokens. PlugClaw deducts credits according to the model used and the amount of processing involved.

A Token is a basic unit used to measure the content a model reads and generates. Text is divided into Tokens before it is processed. Questions, conversation history, prompts, and file content contribute input Tokens, while the model's response contributes output Tokens. Credits provide the balance needed to cover this usage. If the balance is insufficient, tasks that depend on a model may not be completed.

US dollars and credits are converted at a rate of USD 1 = 100 credits, or 1:100. For example, a model call that consumes 200 credits is equivalent to a cost of USD 2.

How Credits Are Consumed

The number of credits used by a task mainly depends on the following factors:

• Model selected: Input, output, and cache-write prices vary by model. Two models may consume different amounts for the same number of Tokens.
• Input size: Questions, conversation history, prompts, and files read during a task all contribute to the input. Longer input usually costs more.
• Output size: Responses, documents, and other generated content contribute to the output. Longer output usually costs more.
• Task process: A complex task may make several model calls or use more context, so it may produce more than one charge.

Model prices usually list separate rates for input, cache writing, and output. Credit usage can be understood as:

Credits consumed
= input Tokens x input credit rate
+ cache-write Tokens x cache-write credit rate
+ output Tokens x output credit rate

When a rate is shown in US dollars per million Tokens, it can be converted using USD 1 = 100 credits:

Credits per million Tokens = USD price per million Tokens x 100
Credits per Token = credits per million Tokens / 1,000,000

Long conversation histories, multi-step tasks, and tool calls may result in several model calls. Actual usage therefore depends on more than the visible length of the final response, and the amount shown in Bill Record is the final charge.

A more capable or more expensive model is not necessary for every task. Everyday questions and short summaries can often use a lower-cost model that meets the task requirements. Complex reasoning, long-context work, and specialized tasks may benefit from a different model.

Model prices change dynamically as services are updated. Open Model Pricing Reference to compare the currently documented rates. Those prices are for reference only; the latest price shown in PlugClaw and the final usage record take precedence.

How to Get Credits

Credits can be purchased from PlugClaw through Stripe:

1. Open the sidebar from the profile entry and go to Recharge.
2. Select a recharge amount and review the number of credits in the order summary.
3. Choose Stripe, accept the User Agreement, and continue with GO TO CHECKOUT.
4. On the Stripe checkout page, verify the merchant and amount, then enter the email address, supported card details, cardholder name, and country or region to complete payment.
5. Return to PlugClaw, check the credit balance, and open Billing to confirm that the order was credited to the account.

Available recharge amounts, the amount charged, and the credits received may change with the payment page. Review the displayed order information before paying.

Purchased credits are valid for one year from the date of purchase. After checkout, review the balance and purchase date so the credits can be used within their validity period.

If the balance does not update immediately, wait briefly and check Billing before attempting another payment. If the card has been charged but the order remains missing, keep the Stripe receipt, payment time, and order details, then report the issue through Settings > Feedback. Do not make a second payment merely to test the first one.

How to Review Credit Usage

Bill Record shows the model calls that consumed credits, including the model used, input quantity, output quantity, and credits charged. These records help identify which models and tasks account for most usage and provide a basis for adjusting model choices or task length.

For the complete payment and record-checking process, see Credits and Billing.