Xberg Core, Pro and Enterprise run the same open-source extraction engine. Pro and Enterprise add what runs around it: durable jobs, storage, search, access control, audit and scale. Both are self-hosted, and documents never pass through Xberg-operated systems.
Documentation: pro.xberg.io
This repository holds documentation only. Xberg Pro and Xberg Enterprise are proprietary software, licensed by written agreement. The documentation is licensed under CC BY 4.0, see LICENSE.
| Core | Pro | Enterprise | |
|---|---|---|---|
| License | MIT | Proprietary | Proprietary |
| Availability | Released | Released | By arrangement: public chart, private images |
| Runs as | Library, CLI, Docker, Helm | Docker Compose or Helm | Helm, on your Kubernetes cluster |
| Scale ceiling | One process | One node | A cluster |
| Price | Free | Per license, by contract | Per license, by contract |
| Support | GitHub, Discord | Email, dedicated Slack channel | Private channel, contractual SLA |
The full comparison is in Core, Pro and Enterprise.
One container plus PostgreSQL, for a single team on a single node.
- Durable job queue and history that survive restarts
- Storage on S3, GCS, Azure Blob or the filesystem, with presigned uploads
- API keys and multiple projects, isolated by row-level security in the database
- Single sign-on through any OIDC provider
- Web console with a playground, job history and settings
- Crawling, content enrichment, and source connectors for Google Drive, Dropbox, OneDrive, SharePoint, Notion, Confluence, S3, GCS, Azure Blob and SFTP
- RAG collections with hybrid search and reranking on PostgreSQL with pgvector
- An opt-in private MCP listener for local MCP clients, off by default (see the MCP guide)
- Auto-Tune, which searches your own corpus for the settings that extract it best
- Append-only, hash-chained audit log
- Prometheus metrics and OpenTelemetry traces
- Python, TypeScript and Go client SDKs
- Monthly job and page quotas per project, enforced with
429 quota_exceeded
A valid license turns on everything Pro does. Nothing is sold as an add-on.
Enterprise runs everything in Pro on your Kubernetes cluster, with the multi-user and multi-tenant layer on top:
- Teams, invitations and roles
- SAML 2.0 single sign-on and SCIM provisioning
- Registered webhook endpoints with retries, delivery history and replay
- Document versions and diff
- Data rooms
- Cross-project usage analytics
- Browser-rendered crawls
- Workers that autoscale on queue depth, down to zero
Audio and video transcription is Core only. Pro and Enterprise reject those inputs.
With Core you build the work around the engine: a durable queue with retries and dead-lettering, storage plumbing, API-key authentication and project scoping, an audit trail, a retrieval layer and a console. Pro replaces that work with a supported product. See What Pro replaces.
The image ghcr.io/xberg-io/xberg-pro is public (linux/amd64 and linux/arm64) and needs a signed license token to start. This sketch shows the shape of the service, not a working install:
services:
pro:
image: ghcr.io/xberg-io/xberg-pro:<version>
environment:
DATABASE_URL: postgres://xberg_app:<password>@postgres:5432/xberg
MAINTENANCE_DATABASE_URL: postgres://xberg_maintenance_login:<password>@postgres:5432/xberg
VECTOR_INDEX_BROKER_DATABASE_URL: postgres://xberg_vector_index_broker_login:<password>@postgres:5432/xberg
VECTOR_INDEX_QUEUE_DATABASE_URL: postgres://xberg_index_worker_login:<password>@postgres:5432/xberg
STORAGE_PROVIDER: fs
STORAGE_BUCKET: xberg
STORAGE_FS_ROOT: /data/xberg
XBERG_ADMIN_KEY: <admin key>
XBERG_LICENSE_KEY: <signed license token>
ports:
- "127.0.0.1:8080:8080" # API
- "127.0.0.1:3000:3000" # web console
volumes:
- xberg-data:/data
volumes:
xberg-data:The database logins come from a PostgreSQL role bootstrap that ships in the Pro chart archive, and the full stack adds the PostgreSQL service and a one-shot bootstrap container. The console sets secure session cookies by default, so sign-in over plain HTTP needs SESSION_COOKIE_SECURE=false.
Follow Install Xberg Pro with Docker Compose for the complete file and first start. For Kubernetes, see the Pro Helm guide and the Enterprise Helm guide.
Both products work the same way: you sign a license agreement, Xberg issues your license, and you run the product. Pricing is per license, by contract. A Pro license token is verified offline and is valid for 180 days, followed by a 30-day grace period. An Enterprise license comes with access to the Enterprise images. Xberg makes no phone-home call and sends no usage beacon. See Pricing and licensing.
To buy a license or arrange a trial, write to contact@xberg.io. Report security issues through the security policy, not a public issue.
- Xberg: the open-source content-intelligence engine behind every tier
- sdks: official client SDKs for the extraction API
- crawlberg: web crawling and scraping with HTML to Markdown conversion
- html-to-markdown: a fast, lossless HTML to Markdown engine
- liter-llm: a universal LLM API client with 170+ providers
- tree-sitter-language-pack: tree-sitter grammars and code-intelligence primitives
- sceptre: EasyOCR's accuracy in a single Rust binary
- alef: the polyglot binding generator