Open-source RAG API written in Rust. Sub-200ms time to first token, 94.5% accuracy on 3,000+ questions. Hybrid search with reranking, streaming answers. Upload PDFs, ask questions, get answers fast. Self-host the whole thing in minutes. Connect knowledge to your Voice AI, AI Agents and any product where retrieval speed is the bottleneck.
Hi everyone, I'm Ignas, the builder behind RustyRAG.
I built this because every RAG solution I tried was either too slow for real-time use cases or too complex to self-host. Voice AI agents, chatbots, and real-time products can't afford to wait seconds for a retrieval pipeline to respond. So I wrote one in Rust that does the whole thing in under 200ms.
I benchmarked it on 3,000+ questions from Open RAG Bench using 1,000 academic PDFs. With reranking enabled, it hits 94.5% accuracy. Without reranking, you get sub-200ms time to first token.
The whole stack is open source. You can self-host it, upload PDFs, and start asking questions in minutes. No GPU required for the server.
I'd love to hear what you think and what use cases you'd want to try it on.
Report
No reviews yetBe the first to leave a review for RustyRAG