How Acadwrite credits are used
See how every feature is billed: conversations by actual usage, and generation or text-processing workflows by document.
Conversations · Actual usage
Postpaid. Start with any positive balance and pay for the tokens actually used when the conversation ends.
Literature search, citation annotation, reviews, writing, and other work performed in a workspace conversation all count toward that conversation's token usage.
Example: a typical question uses about 1–2 credits; completing a literature review in a conversation uses about 8–9 credits.
Generation · Per document
A fixed prepaid amount, independent of document length, is reserved before the workflow begins.
Turn a topic into a structured first draft with real citations.
Generate a presentation outline and slide-by-slide content.
Generate a paper introduction around a research topic.
Receive a rigorous, detailed review with annotations.
Search across sources and organize results by research direction.
Text processing · Per document
A fixed prepaid amount independent of text length. Credits are charged only on success and returned automatically on failure.
Rewrite text academically to reduce signals of AI generation.
Find and annotate real literature citations for a passage.
Improve language and academic expression.
Rewrite text to reduce similarity with existing literature.
Frequently asked questions
How is credit usage calculated?
Knowledge Assistant uses actual token consumption. Generation workflows such as reviews, slides, and introductions have a fixed per-document charge. Text-processing workflows such as AI-score reduction, paraphrasing, editing, and citation matching also have a fixed per-document charge. See the table above for exact amounts.
Why are conversations billed by token instead of per document?
Conversation workloads vary with search depth, response length, and invoked subtasks. Usage-based billing keeps simple questions inexpensive while charging more only for complex work.
How many credits does a literature review use?
The standalone literature-review workflow costs 15 credits per document, regardless of length. A review completed inside Knowledge Assistant is included in the conversation's actual token usage instead.
What is a token, and how does it relate to word count?
A token is the smallest text-processing unit used by an AI model. As a rough estimate, one Chinese character is about 1.5 tokens and one English word is about 1.3 tokens.
What happens when I do not have enough credits?
Prepaid workflows check your balance before starting and ask you to add credits if needed. A conversation can start with any positive balance and is settled after it ends.