Extraction Pricing
No subscriptions. You buy credits once and spend them as you go. This page explains exactly what causes credits to be deducted.
How the credit system works
Credits are the currency of the extraction engine. When you purchase credits, they are added to your account balance. Every successful extraction job deducts credits based on the job\'s complexity. Credits never expire.
New accounts receive a free credit allocation at sign-up, so you can run real jobs before deciding whether to top up.
What determines the cost
1. Number of pages scanned
The AI must read every page to decide whether to keep it. Uploading a 200-page textbook costs more than uploading a 20-page paper. This is the dominant cost factor for most jobs.
2. Prompt complexity
Simple prompts ("keep questions about momentum") use less compute than bucket jobs with 5 categories and complex exclusion rules. More buckets = slightly higher base cost.
3. Output mode
Markdown mode costs slightly more than PDF mode because the AI re-generates each kept page as clean structured text rather than just copying raw page data.
Example job costs
| Scenario | Pages | Mode | Approx. cost |
|---|---|---|---|
| Single past paper, 1 topic filter | 22 | Low | |
| 3 papers merged, 2 bucket split | 66 | Medium | |
| 150-page lecture deck, formulas only | 150 | Medium | |
| 400-page textbook, 5-way bucket split | 400 | Higher | |
| 20-page paper, Markdown output | 20 | Markdown | Low |
| 80-page notes, Markdown output | 80 | Markdown | Medium |
No charge for failures
Credits are only deducted when a job completes successfully and produces a download. If the job fails for any reason — corrupted file, server error, timeout, or a bug on our end — no credits are deducted and the balance is fully refunded to your account.