LlamaCloud

LlamaCloud is LlamaIndex's managed platform for the document side of RAG, distinct from the open-source framework. Three products cover the pipeline: LlamaParse, a GenAI-native parser that converts complex PDFs, DOCX, PPTX, and spreadsheets across 130+ formats into LLM-ready Markdown or JSON; LlamaExtract, which turns documents into typed JSON against a Pydantic or JSON schema; and LlamaCloud Index, a fully managed ingestion, embedding, and hybrid-retrieval endpoint with reranking.

In September 2026 LlamaIndex CEO Jerry Liu signaled the company's strategic bet: the primary focus shifted to LlamaParse, the newer LiteParse, and the ParseBench/ExtractBench benchmarks, with the open-source framework continuing as a toolkit. Developers should note the SDK transition: the classic llama-cloud-services v1 API is legacy and maintained only through May 1, 2026, while the new llama-cloud v2 SDK is the current path for new projects. The two APIs are not drop-in compatible.

For teams whose RAG quality lives or dies on parse quality, managed parsing beats maintaining OCR and layout pipelines. The risk is platform concentration: the vendor's center of gravity is moving toward parsing products, and retrieval-heavy roadmaps should account for the slowing open-source framework cadence.

Sign up at cloud.llamaindex.ai, get an API key, use the llama-cloud v2 SDK or REST API.