LlamaParse

LlamaParse is the managed document-parsing service inside LlamaCloud, LlamaIndex's enterprise agentic AI platform. It handles 130+ document formats with agentic OCR, extracting text, tables, charts, and images into high-quality markdown optimized for RAG and LLM ingestion, with support for custom parsing instructions like table-as-CSV rules or header/footer exclusion. Beyond parsing, the platform includes Extract (structured data extraction), Index (ingest and RAG pipelines), and LlamaAgents (deployed document agents).

It has processed over a billion documents for more than 300,000 users across finance, insurance, healthcare, legal, and manufacturing, and topped its own ParseBench benchmark at 84.9% overall in Agentic mode. The older llama-parse Python package is deprecated in favor of the unified llama-cloud SDK. Limitations: ParseBench is LlamaIndex's own benchmark so the headline score is self-reported, it is a paid cloud API with usage-based pricing (not a local library), and high-stakes pipelines should cross-reference accuracy against independent evaluations.

Create a LlamaCloud account at cloud.llamaindex.ai, export LLAMA_CLOUD_API_KEY, and install the SDK: pip install llama-cloud or npm install @llamaindex/llama-cloud.