Skip to content
FastRAG
Release history

Changelog

Follow our journey building the ultimate RAG Starter Kit. We ship meaningful updates every few weeks.

Get FastRAG
  1. v2.0.0Latest

    Claude Haiku · Voyage AI · Cheerio

    A full stack swap — replaced OpenAI with Anthropic Claude Haiku for chat, Voyage AI voyage-3.5 for embeddings (200M free tokens), and Cheerio for serverless-safe web scraping. Zero ESM conflicts on Vercel, no headless browser required.

    • NewClaude Haiku (claude-haiku-4-5) — sub-second TTFT streaming via native Anthropic SDK SSE
    • NewVoyage AI voyage-3.5 embeddings — separate document and query embedders for asymmetric retrieval
    • NewCheerio web scraping — fully CommonJS, no Puppeteer or Browserless required
    • NewIP-based demo rate limiting — 5 messages + 3 ingestions per 24h, per visitor
    • NewDemo status API — /api/demo-status returns usage counters and resetAt for live UI meters
    • NewFirst SSE event carries sources + usage payload before any answer text begins
    • ImprovedPDF ingestion increased to 20MB per file via Formidable multipart upload
    • ImprovedPinecone namespace isolation — namespace flows uniformly through ingest → store → retrieval
    • Improved0.3 cosine similarity threshold filters irrelevant chunks before Claude is called
    • FixRemoved BROWSERLESS_TOKEN and OPENAI_API_KEY — replaced with ANTHROPIC_API_KEY + VOYAGEAI_API_KEY
    • FixResolved ERR_REQUIRE_ESM on Vercel caused by node-html-parser — replaced with Cheerio
  2. v1.3.0

    The Production Overhaul

    A complete rewrite of the ingestion engine and UI. Moved from a basic script interface to a production-grade SaaS layout with multi-user support.

    • NewWeb Scraping Engine (Puppeteer + Browserless)
    • NewBento Grid Landing Page & full UI refresh
    • NewMobile-first Chat Interface
    • ImprovedInteractive Terminal Loader for demos
    • FixNamespace Isolation — prevents context bleeding between users
  3. v1.2.0

    PDF & Cost Optimisation

    Focused on reducing vector storage costs and handling larger documents more reliably.

    • NewMulti-file PDF Upload Support (up to 10 files)
    • ImprovedMigrated to Pinecone Serverless for lower costs
    • FixFixed PDF parsing for large files (>10 MB)
  4. v1.0.0

    Initial Launch

    The first public release of the FastRAG Starter Kit.

    • NewNext.js Pages Router setup
    • NewBasic RAG pipeline — LangChain + OpenAI
    • NewTailwind CSS styling + Vercel deploy config

Ready to build AI that knows your data?

Grab the source, wire up your API keys, and have a production RAG app running today.

Start buildingTry FastRAGExplore docs

Pay once · MIT license · Lifetime updates