Dify
An open-source platform for developing LLM applications, combining visual AI workflows, RAG pipelines, agents, model management and observability, deployable in the cloud or self-hosted.
Dify is an LLM app development platform. Workflows are built and tested on a visual canvas, and a Prompt IDE compares prompts and model performance.
It supports hundreds of proprietary and open-source models through many inference providers and any OpenAI-compatible API. Its RAG pipeline covers document ingestion and retrieval, including text extraction from PDFs and slides. Agents run in their own sandbox and can use skills, marketplace tools, MCP servers and custom APIs. LLMOps features log and analyze application performance, and every feature has an API.
Dify runs as a hosted cloud service or self-hosted with Docker Compose. It is written in TypeScript and Python.