LlamaParse

What it does

LlamaParse takes in PDFs, scans, Office files, spreadsheets and images, including pages with dense tables, charts, handwriting and checkboxes. It returns markdown, plain text or JSON per page, with options such as annotated PDFs and bounding boxes. Separate Extract, Classify, Split and Index functions return schema-shaped JSON, sort documents by natural-language rules, break combined files into sections, or prepare content for retrieval.

Developers call it through one API key with Python, TypeScript, Go and Java SDKs, or try it in a web console. The distinguishing feature is tiered parsing (Cost Effective, Agentic, Agentic Plus) so users can trade cost against accuracy, plus an Auto Mode that routes tiers page by page. Extraction can include citations and confidence scores. It is a commercial hosted platform, not the open-source LlamaIndex framework; a separate open-source local parser, LiteParse, also exists.

It suits engineers and teams building document pipelines or retrieval systems, particularly in finance, insurance, healthcare and manufacturing. Use requires signing up at the cloud console and an API key. Usage is credit-based, so costs depend on the mode chosen and should be tested on your own document types. Enterprise options include VPC deployment, SSO and compliance certifications.

What it can do

• Parse PDFs, scans, Office files and images to markdown, text, JSON or XLSX
• Layout-aware table, chart and handwriting parsing with bounding boxes
• Schema-based Extract with citations and confidence scores
• Classify documents by natural-language rules and Split combined files
• Managed Index for ingestion, embedding and retrieval
• Tiered parse modes and Auto Mode for per-page cost control

Example workflows

Convert scanned invoices to structured data

Input: multi-page scanned invoice PDFs. Steps: upload with the parse purpose, parse with the agentic tier, then run Extract with a defined schema. Output: markdown plus JSON fields with citations back to pages.

Prepare reports for retrieval

Input: a combined PDF of reports. Steps: Split into logical sections, Parse tables and charts, add to an Index. Output: a managed index queryable for retrieval-augmented answers.

Before you choose

• Credit cost varies by parse tier and options, so test settings on your own documents before building pipelines
• Cloud SaaS caches data for 48 hours by default (can be turned off); VPC deployment is an enterprise option
• LlamaParse is commercial and not open source; LiteParse is the separate open-source local option

Listed under

• Data
• Data extraction
• Developers
• Document parsing
• Enterprise
• OCR
• PDF

Get started

Create an account; API key needed for SDK use

1. Sign up at cloud.llamaindex.ai and create an API key
2. Set the LLAMA_CLOUD_API_KEY environment variable and install an SDK
3. Run a sample parse, then test tiers on your own documents

Credit-based; each action costs credits depending on mode. Starter and Pro allow pay-as-you-go up to $500/mo and $5,000/mo. Enterprise is custom. Recurring monthly plan fees not shown in the product information.

Source: llamaindex.ai/pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• LlamaIndex | AI Agents for Document OCR + Workflows ↗www.llamaindex.ai
• AI Parsing Tool for AI-Ready Document Data | LlamaIndex ↗www.llamaindex.ai
• LlamaParse Pricing: Compare Plans & Credits | LlamaIndex ↗www.llamaindex.ai
• LlamaParse documentation | Developer Documentation ↗developers.llamaindex.ai

Company website

llamaindex.ai: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.