docling-project/docling.rs

Set these on your host before deploying. Each platform has an “Environment Variables” section in the deploy flow. (showing first 20, more exist in .env.example)

• RAG_DB_BACKEND Database (vector store) sqlite | postgres | memory
• RAG_DATABASE_URL secret sqlite://<path> or postgres://user:pass@host:5432/dbname
• RAG_EMBED_PROVIDER Embeddings ollama | gemini | onnx | hash
• RAG_EMBED_MODEL
• RAG_EMBED_DIM
• RAG_CHUNK_UNIT word | token
• OLLAMA_BASE_URL Ollama (default provider)
• GEMINI_API_KEY secret Gemini (RAG_EMBED_PROVIDER=gemini)
• RAG_GEMINI_MODEL
• RAG_EMBED_ONNX_PATH Local ONNX (RAG_EMBED_PROVIDER=onnx; build with --features onnx-embed). Fetch the bge-m3 model into these default paths with:…
• RAG_EMBED_TOKENIZER secret
• RAG_OCR_LANG OCR recognition language for scanned PDFs: ch (default, multilingual, conformance-validated) or en (English-only PP-OCRv3 — much better…
• OPENROUTER_API_KEY secret LLM (OpenRouter) — for Multi-Query, HyDE, and answer synthesis OpenRouter keys start with "sk-or-". Any OpenAI-compatible endpoint works:…
• OPENROUTER_BASE_URL
• RAG_LLM_MODEL
• RAG_CHUNKER Chunking window (streaming Markdown sliding window, default) | hierarchical | hybrid (docling's structure-aware chunkers; hybrid uses…
• RAG_CHUNK_SIZE
• RAG_CHUNK_OVERLAP
• RAG_RETRIEVAL_MODE Retrieval vector | bm25 | hybrid | multi-query | hyde
• RAG_TOP_K