LlamaParse
A hosted document parsing service from the LlamaIndex team for hard documents: scanned PDFs, forms, tables, charts, slides and spreadsheets, across 130+ formats. It returns Markdown, text or JSON, and the same API key covers extraction to a schema, classification and managed indexing. Billed in credits, with a free monthly allowance.
10K credits a month
llamaindex.ai/pricing · Oct 2026
Tool
AI & Machine Learning
LlamaIndex
An hour to try, then reference
Intermediate
Free 10K credits a month; paid plans
What LlamaParse is
LlamaParse is a hosted platform for turning difficult documents into clean, model-ready text. It comes from the team behind the LlamaIndex framework, but works with any stack through its API and SDKs for Python, TypeScript, Go and Java.
What it offers
Several products share one API key:
- Parse: converts documents to Markdown, text or JSON, with what the docs call agentic OCR for scans, tables, charts and forms across more than 130 formats.
- Extract: pulls typed data that matches a schema you define, with citations back to the source.
- Classify: sorts documents by type using natural-language rules.
- Split: breaks a combined file, such as a scanned bundle, into its separate documents.
- Index: managed ingestion and retrieval for RAG.
Parsing has modes from basic, cheap parsing to layout-aware parsing that uses language or vision models for higher accuracy at a higher credit cost.
When to choose it
A hosted parser makes sense when your documents are hard (multi-page tables, charts, handwriting, poor scans), you have no time to tune an open-source pipeline, and your data is allowed to leave your servers. If documents are sensitive or must stay in a country, start with a local tool such as Docling instead, or confirm the provider's data terms and regions first.
Whichever you choose, test on 20 of your ugliest real documents and read the output yourself. Parsing errors silently become wrong answers later.
Prerequisites and time
An account, an API key and basic Python or TypeScript. A first parse takes minutes.
Free or paid?
As of October 2026 the free plan includes 10,000 credits a month; paid plans start at a monthly fee and add more credits and features. Credits per page depend on the parsing mode, so check the pricing page before processing large batches. The team also publishes LiteParse, an open-source option for local parsing.
Common questions
LlamaParse or Docling? Docling runs locally and is free; LlamaParse is hosted and aims at the hardest layouts. Test both on the same 20 documents and compare the output by eye before you decide.
Best for: projects over messy PDFs, scans and spreadsheets where parsing quality matters more than keeping everything local.
Ready to start?
Opens on LlamaIndex in a new tab.
Stuck on something specific?
Writing only gets you so far. If you want an answer to your situation rather than the general case, book a session and we will work through it together. Sessions are free for approved Sefism members, and a few slots open each week.
Follow along
New writing, resources and project ideas land here first.