Xberg
Open-source polyglot document intelligence framework with Rust core for 96+ formats
Best open-source choice for CPU-efficient document extraction at scale, with unmatched polyglot SDK support. The v5.0 ML additions widen the gap versus Unstructured. Pass if you need a no-code GUI or prefer GPU-native OCR from Azure Document Intelligence.
- Developers building high-throughput document extraction pipelines for RAG or data lakes
- Teams needing polyglot SDK support (Python, TypeScript, Rust, Go, etc.) in microservices
- DevOps deploying CPU-only extraction in containers or serverless environments
- Researchers processing academic formats like LaTeX, BibTeX, and JATS
- Non-technical users: requires programming knowledge or CLI usage
- GPU-accelerated OCR at scale: designed for CPU efficiency; some VLM OCR may need GPU
- Real-time speech transcription: Whisper is file-based, not streaming
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
In short
Xberg — Open-source polyglot document intelligence framework with Rust core for 96+ formats. Best for Developers building high-throughput document extraction pipelines for RAG or data lakes, Teams needing polyglot SDK support (Python, TypeScript, Rust, Go, etc.) in microservices, DevOps deploying CPU-only extraction in containers or serverless environments. Free to use.
Viability Score
How likely is Xberg to still be operational in 12 months? Based on 4 signals — momentum (how recently it shipped), wrapper dependency, revenue model, and web presence.
Last calculated: July 2026
How we score →Key Features
- Extract text, tables, metadata, images from 96+ formats
- Multi-engine OCR: Tesseract, PaddleOCR, EasyOCR, VLM
- Whisper ONNX audio/video transcription (v5.0)
- Code intelligence: functions, classes, imports from 306 languages
- Named entity recognition and redaction (v5.0+)
- Document summarization and translation (v5.0+)
- Page classification and VLM image captions
- QR-code detection
- Plugin system: custom extractors, OCR backends, validators
- REST API, CLI, Docker, MCP server deployment
- Native SDKs in 17 languages (Python, TS, Rust, Go, Java, etc.)
- SIMD-optimized CPU pipeline (no GPU required)
- PDF page rendering (v4.6+)
- Image-index references and SVG normalization (v5.0)
- HEIC aggregate format support (v5.0)
About Xberg
Kreuzberg is an open-source document intelligence framework built on a Rust core that extracts text, tables, metadata, and audio transcripts from 96+ file formats. It serves developers building scalable document processing pipelines for tasks like RAG, data extraction, and archiving. The framework offers native SDKs in 17 programming languages, including Python, TypeScript, Rust, Go, Java, and C#, and deploys as a library, CLI, REST API, Docker image, or MCP server. Key features include multi-engine OCR (Tesseract, PaddleOCR, EasyOCR, and VLM-based via Candle), optional Whisper ONNX audio/video transcription (v5.0+), code intelligence from 306 programming languages, and a plugin system for custom extractors and backends. The v5.0 release adds image-index references, SVG/image normalization, HEIC support, and advanced ML features like named entity recognition, redaction, summarization, and translation. Kreuzberg's Rust core eliminates dependencies on external tools like LibreOffice, delivering high throughput (thousands of documents per minute) without a GPU, with SIMD optimizations and parallelism. The multi-backend OCR pipeline includes document-level optimization. Integrations with LangChain, LlamaIndex, Haystack, and others extend its reach into AI workflows. Compared to Unstructured or Apache Tika, Kreuzberg offers faster CPU-only performance, broader language support, and a more flexible plugin architecture, though its feature richness may overwhelm newcomers and some capabilities remain experimental.
Behind the Verdict
Kreuzberg is the most language-agnostic document extraction tool we've seen. Its 17 SDKs are a genuine differentiator — teams mixing Python, TypeScript, and Go can all use the same core library. The Rust engine is fast and dependency-light, ideal for containers and serverless. We'd reach for this when building a document pipeline that must process diverse formats (PDFs, Office, images, emails) at high throughput without a GPU. It's a natural fit for RAG backends where latency and cost matter. The new v5.0 ML features (NER, summarization, translation) add real value, though they are experimental. Where it bites: OCR quality trails dedicated engines like Azure Document Intelligence for scanned documents. Some advanced features are only available in Python (EasyOCR) or require Wasm/ONNX builds. The configuration surface is deep — expect a learning curve. Compared to Unstructured, Kreuzberg is faster on CPU and has broader SDK coverage, but Unstructured offers a cloud API and more polished PDF extraction. For pure speed and polyglot support, Kreuzberg wins. For turnkey cloud, choose Unstructured. In practice, Kreuzberg shines in microservices where each service needs document extraction in its native language. The plugin system is powerful but underdocumented — expect to dig into source. We recommend starting with the Python or TypeScript SDK for fast prototyping.
Researching Xberg? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Use Cases
- Extract full text and tables from PDF bank statements for automated data entry
- Parse academic papers in LaTeX, BibTeX, or JATS formats for research indexing
- Process employee onboarding documents (Word, Excel, ID scans) through a unified pipeline
- Transcribe meeting audio files and extract action items via Whisper + OCR
- Classify incoming email attachments and route them based on content type
- Build a RAG pipeline by chunking and extracting metadata from 96+ file types
Models Under the Hood
as of 2026-07-17
Limitations
- Xberg is free and open source with no rate limits, but OCR performance depends on the chosen engine.
- Some advanced features like VLM-based OCR require a separate LLM/vLM backend.
- The WASM build is limited to ~60-80% native speed and cannot use libheif for HEIC images.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Integrations
Resources & Guides
Official links
Tools that pair well with Xberg
Common stack mates teams adopt alongside Xberg, with the specific reason each pairing earns its keep.
Featured Head-to-Head Comparisons
Alternatives to Xberg
View all
Arize Phoenix
Open-source AI observability for LLM agent tracing and evaluation.
OpenAgents
Open-source platform for deploying language agents in everyday scenarios.
Frequently Asked Questions
Best-of guides
Used Xberg? Help shape our editorial sentiment research.