Guides

Learn how to build RAG applications with the Neutron API

How Neutron Works

Ingestion
Upload and process your documents
Document/Text
PDF, Text, Markdown
Seed
Content record
Chunks + Embeddings
~300 tokens, 1024-dim vectors
Vector DB
pgvector storage
Query (RAG)
Search and generate responses
Question
Natural language
Embedding
Query vectorization
Similarity Search
Find relevant chunks
Context
Merged results
LLM Response
GPT-4, Claude, etc.

Quick Reference

Base API URL

https://api-neutron.vanarchain.com

Authentication

Authorization: Bearer nk_...