comparisons
LangChain vs LlamaIndex: Production Verdict for 2026
LangChain vs LlamaIndex in 2026: LangChain for agent loops, LlamaIndex for retrieval-heavy RAG. Versioning, t…
LlamaIndex Hub — Build and debug RAG retrieval pipelines
Build and debug RAG retrieval pipelines
LlamaIndex Hub
Independent · 2026
Independent · 2026
fundamentals
LlamaIndex VectorStoreIndex Explained for Production RAG
How VectorStoreIndex embeds nodes, where text lives, why similarity_top_k defaults to 2, and how to persist, filter and update it in production.
Read the article →Latest guides
llamaindex
LlamaIndex Query Engine vs Chat Engine for Production
Choose the right LlamaIndex interface by comparing state, retrieval behavior, latency, evaluation, and deploy…
how-to
LlamaIndex Chunk Size: Tune chunk_size and chunk_overlap
Configure LlamaIndex chunk_size and chunk_overlap with SentenceSplitter, per-index transformations, and a rep…
comparisons
LlamaIndex vs LangChain: Which to Use for RAG
A documentation-based comparison of LlamaIndex and LangChain for retrieval: what each project optimises for, …
troubleshooting
LlamaIndex Retrieval Troubleshooting: Fix Bad Answers
Nine failure modes in LlamaIndex retrieval, how to tell them apart from source nodes and scores, and the docu…
tutorials
LlamaIndex Quickstart: Build a RAG Pipeline in Python
Install LlamaIndex, index a folder of documents, query it, persist the index, and swap in a real vector store…
Start here
A reading order for LlamaIndex retrieval, from what the pipeline is doing to what to change when the answers come back wrong.
- Start with the conceptsLlamaIndex Ingestion, Indexing and Retrieval Explained
How documents become nodes, and which index type answers which kind of question.
- Then build oneLlamaIndex Quickstart: Build a RAG Pipeline in Python
Install, index a folder, query it, persist it, and move to a real vector store.
- Check the framework choiceLlamaIndex vs LangChain: Which to Use for RAG
What LlamaIndex and LangChain each optimise for, and why the switch is cheaper than it feels.
- Fix it when answers are wrongLlamaIndex Retrieval Troubleshooting: Fix Bad Answers
Nine retrieval failure modes, told apart from the source nodes and scores.
Sizing an index before you provision anything? Use the chunking and vector store sizer to turn corpus size, chunk size and embedding dimensions into a chunk count and a RAM figure.