PDF Parsing Modes

Firecrawl supports three PDF parsing modes. The core tradeoff is speed vs. robustness across different PDF types.

Picking the right mode

Mode	Best for	Strength	Tradeoff
fast	Text-native PDFs	Fast	Won’t extract from scanned/image-heavy pages
auto	Default for most PDFs	Text-first with OCR fallback	More complex but more robust for mixed PDFs
ocr	Scans, photos, and auto misclassification	Most robust	Higher cost and latency

Example: force OCR

parsers: [{ type: 'pdf', mode: 'ocr', maxPages: 20 }]

Example: default behavior

parsers: [{ type: 'pdf' }]

Practical guidance

• Start with auto and switch to ocr only for the PDFs that need it
• Use fast when you know the PDFs have a clean embedded text layer
• For batch pipelines, sample your PDF distribution first and then choose defaults

Related:

• Scanned PDFs & OCR
• Performance & Cost