RAGFlow on Railway
Retrieval-augmented generation engine with deep document parsing and cited answers, on Railway.
RAGFlow is an open-source retrieval-augmented generation (RAG) engine built around deep document understanding. It parses PDFs, scans, tables and slides into structured chunks and answers questions with citations that point to the source passages.
Teams use it to chat with contracts, policies, manuals and research papers, to build knowledge bases over documents that plain text extractors handle poorly, and to expose a retrieval API or MCP endpoint for AI agents. It also has an agent workflow builder that combines retrieval with web search and tools.
RAGFlow is licensed Apache-2.0.
What RAGFlow can do
- Deep document parsing with OCR and layout models
- Chat with cited source passages
- Knowledge bases over PDFs, scans, tables and slides
- Retrieval API and MCP endpoint
- Agent workflows with retrieval, web search and tools
- Chat and embedding models configured in the UI
What people use it for
- Chat with contracts, policies, manuals and papers
- An internal knowledge base over scanned PDFs and tables
- A retrieval API for AI agents
- Agent workflows that mix retrieval and web search
Videos
Running RAGFlow on Railway
The Railway template deploys RAGFlow v0.27 as a web service and a separate parsing worker, with Elasticsearch 8.11.3, MySQL 9 and Redis 8.2 on volumes, and a Railway Bucket for uploaded files through RAGFlow's S3 backend instead of MinIO. Passwords and secrets are generated. It is a memory-heavy stack: the template's planning estimate is roughly 6 to 10 GB of RAM in total, an estimate and not a measurement. Parsing capacity scales with worker replicas.
RAGFlow templates on Railway
Tracklayer templates are listed first. Other RAGFlow templates follow, ordered by total deployments on Railway. How this works.
Deployment numbers as of 2026-11-20.
Questions about RAGFlow
How do I log in?
Wait for the ragflow healthcheck to pass, which can take several minutes on first start, open the public URL and register. Registration is open by default, so set REGISTER_ENABLED to 0 on the ragflow service afterwards.
Where do model keys go?
In the UI. Add your chat and embedding model providers, such as OpenAI, Azure, Gemini, Ollama or a LiteLLM gateway.
What drives the cost?
Memory: Elasticsearch, the parsing worker (each executor needs roughly 1.5 to 3 GB while parsing) and the web service. The template estimates 6 to 10 GB of RAM in total.
Is my data kept across redeploys?
Yes. Elasticsearch, MySQL and Redis use volumes, and original files are stored in the Railway Bucket.
Are there licence notes?
RAGFlow is Apache-2.0. Elasticsearch 8.11 is distributed under Elastic License 2.0 or SSPL, and running it unmodified as a backing service is permitted under ELv2. MySQL Community is GPL-2.0 and Redis 8 is RSALv2, SSPLv1 or AGPLv3.
Guides
Similar apps
Tracklayer maintains the templates marked as such; RAGFlow is a trademark of its owners and this site is not affiliated with them.