Understanding Quotas & Intelligence Units

Storage Capacity

Storage capacity works like traditional cloud storage (e.g., Google Drive). It's a persistent quota that:

Intelligence units are consumed when processing files for indexing. The consumption rate varies by file type:

PDFs with OCR: 1 unit per 10 pages
OCR is automatically applied when text extraction fails
Normal PDFs: 0 units
When text can be extracted directly
Other Files: 1 unit per 5 pages
Including DOCX, PPTX, and other supported formats
Audio & Video: 1 unit per minute
Automatic transcription with speaker diarization. Max 1 GB file size, 10 hours duration per file. Transcript searchable with timestamps.

Native Page Formats:
PDF, PPTX, TIFF: Direct page/slide count
DOCX:
Uses metadata page count when available
Other Formats:
Calculated as: file size ÷ 100KB per page
Audio & Video:
Page count = Number of transcript chunks (~12,000 chars each). Each chunk represents a continuous segment of speech with precise timestamps.

Translating PDF files is an Enterprise-only feature that consumes intelligence units based on the number of pages:

PDF Translation: 4 units per page
Every page in the PDF consumes 4 intelligence units when translated. For example, a 50-page document would consume 200 intelligence units.
Translation preserves document formatting, structure, and layout while converting the text to your target language using Google Cloud Translation API. Supports both native PDFs (with selectable text) and scanned PDFs.

When asking questions, intelligence units are also consumed based on:

Complexity of the question
Amount of context needed to answer
Number of files being analyzed
Processing depth required (simple lookup vs. complex analysis)
Citation Formatting:APA citation generation (reference list or in-line) consumes intelligence units based on the number of sources and complexity

Indexed pages represent the number of pages processed and made searchable in your account: