Understanding AI Pricing | Notion

<aside>

On this page:

</aside>

<aside>

Jump to video ↓

</aside>

<aside>

🧠 Product Docs

Expanse 101
Tips & Tricks

🤖 Prompting Tips

Prompting Tips & Tricks

💻 Use Cases

📚 Resources

Prompt Engineering Resources

🍄 About Expanse

</aside>

Why does it cost money to use AI?

Every time you chat with AI, it uses real computing power and resources. Here are some examples of what the cost of AI covers:

Uses electricity to power big computers and servers which store the data
Needs regular maintenance
Requires ongoing “training” to keep the models up-to-date

How the Cost of AI is Calculated

The cost of your prompting to an AI is generally calculated in Tokens, which are a basic unit of text that the AI models process.

Think of tokens as "AI words"
Each token is about 4 characters (letters, spaces, or symbols)
Example: "Hello there" = about 3 tokens

Why do some models cost more than others?

The cost of AI models varies based on several factors:

Model Size: Larger models with more parameters require more computing power
Capabilities: More advanced features like code generation or image analysis need more resources
Speed: Faster response times require more powerful infrastructure
Quality: Higher-quality outputs often come from more sophisticated models
Training Data: Models trained on larger datasets typically cost more to develop and maintain

You can view the cost of models in Expanse by going to the Model Pricing tab in Settings.

Common Terms & Abbreviations

Term	Definition
Token	The individual units of text that AI models break down and process. Eg. Words, punctuation marks, and parts of words. 4 English characters ≈ 1 token.
Tokens I/O	Tokens In/Out: The number of tokens in the user’s message sent to the AI (input), and the number of tokens the AI generates in its response (output).
Input MTok	The token cost for messages sent to the AI model. Larger messages (like file uploads) use more input tokens.
Output MTok	The token cost for responses generated by the AI model. Longer, more detailed responses use more output tokens.
MTok	A billing measurement where 1 MTok = 1 million tokens. Used to calculate large usage costs for AI interactions.
Context window	The maximum number of tokens an AI model can process and reference at once (including both input and output text).
Maximum Output	The maximum length (in tokens) that an AI model can generate in a single response
Per Image Pricing	The cost for an image to be generated.
Model	A specific version of AI with its own set of capabilities, costs, and performance characteristics.
Provider	An organization that develops and supplies AI models. Each provider can offer multiple AI models with different capabilities and specializations.

Related Docs

Context Management Guide

Early Access Memberships

Didn’t find what you were looking for?

**Expanse.com**