Skip to main content

BookWidgets AI Credits

BookWidgets AI Credits measure the underlying cloud computing power ("tokens") required to process your requests and build interactive widgets. Because our AI acts as an intelligent assistant that reads context, uses specialized educational tools, and formats interactive exercises behind the scenes, credit usage is metered based on total computational tokens processed, rather than just the visible words you type on the screen. To keep metering accurate for both small and large actions, usage is counted in millicredits (mC), where 1 Credit = 1,000 millicredits.

1. What Counts Toward Token Usage?

Every time you send a message, analyze a resource, or generate content, usage is calculated based on the following components:

  • Total Input Tokens (What the AI Reads): This includes the instructions you type, any documents, images, or PDFs you attach, the ongoing conversation history (so the AI remembers your previous messages in the chat), and the behind-the-scenes system instructions and educational formatting rules BookWidgets sends to the AI so it knows how to build valid widgets.
  • Cached Input Tokens (Automatically Reused Context): When repeated prompt context—such as system instructions, earlier chat messages, or previously read text, images, and documents—is automatically served from cache during an active session, those cached input tokens are billed at a reduced cached input rate.
  • Total Output Tokens (What the AI Does & Writes): This includes the visible text, feedback, widget content, or images generated for you, plus any internal "thinking" steps and automated tool calls (such as configuring widget settings, or validating question formats) that the AI performs in the background before showing your final result.

Tip: Longer conversations and multiple regeneration steps require the AI to re-read the earlier chat history and execute multiple tool steps. Starting a fresh chat for a new lesson or widget helps minimize credit usage!

Tip: Processing long documents will consume a lot of tokens, and can become the main factor determining the total credit usage. Only provide the documentation (e.g chapter) required for the task.

2. Standard Credit Conversion Rates

Usage is deducted automatically upon successful completion of an AI action according to the following metered rates (1 Credit = 1,000 millicredits (mC); 1,000 text tokens equal roughly 750 words of raw data):

AI Model & Role Input Rate (Uncached) Cached Input Rate Output Rate (Reasoning, Tool Calls & Final Content) Typical Teacher Action (Estimated Total)
Reasoning AI v1
Interactive AI Assistant (widget creation & editing, reasoning, tool calls, spatial image analysis)
150 mC per 1,000 tokens
(300 mC from January 2027)
(text, image & document/PDF)
15 mC per 1,000 tokens
(30 mC from January 2027)
(cached text, image & document/PDF)
750 mC per 1,000 tokens
(1,500 mC from January 2027)
~2–5 Credits per short chat turn (~4–10 from Jan 2027);
~10–25 Credits per complex widget operation (~20–50 from Jan 2027)
Standard AI v1
Fast Analysis & Support AI (document/PDF reading & indexing, standard/OCR image analysis, context summarization)
50 mC per 1,000 tokens
(text, image & document/PDF)
6 mC per 1,000 tokens
(cached text, image & document/PDF)
300 mC per 1,000 tokens ~0.2–2 Credits per document/image analysis
Image generation AI v1
Image Generation & Modification
300 mC per 1,000 tokens
(text prompt & reference image)
30 mC per 1,000 tokens
(cached text & image)
3,750 mC per 1,000 image output tokens;
1,500 mC per 1,000 text/thinking tokens
~6 Credits per generated or modified image

(Note: Because our AI Credit rates reflect underlying third-party AI compute and token costs, we may introduce rates for new models at any time and may increase existing credit rates for future usage with reasonable prior notice if provider pricing or infrastructure expenses rise, as detailed in our Terms of Service.)