Dgraph v25 is now in preview: namespaces, MCP servers, and no license key required

Dgraph v25 is now available in public preview. We are excited for this launch as this isn’t just a version release—it’s a fundamental shift in how graph infrastructure can be accessed, scaled, and applied to AI-native workloads.

Starting with version 25, all previously enterprise features of Dgraph are freely available, running on modern Hypermode infrastructure. Whether you’re building agent context, managing multi-tenant data, or operationalizing change data capture, Dgraph v25 is ready out of the box—with no license key, no deployment complexity, and no feature gating.

This release reflects a broader commitment from Hypermode: to make high-performance graph systems accessible to every team building context-aware apps, without artificial limitations. Whether you're a solo developer or part of a large platform engineering group, Dgraph v25 gives you production-ready primitives for building graph-native systems, embedded directly in the Hypermode platform.

If you missed our vision post, we encourage you to read: The future of Dgraph is open, serverless, and AI-ready. Today’s preview brings that vision to life.

Why this release matters

Graph databases are no longer niche infrastructure for niche use cases. They’ve become core to how we represent memory, context, relationships, and evolving knowledge inside AI-native systems. From agent toolchains and RAG (retrieval-augmented generation) memory to fraud graphs and personalization engines, the structure of knowledge is now the architecture of intelligence.

But traditional enterprise licensing has held back adoption:

• You needed a special license for core features like access control or audit logging

• Multi-tenant deployments required complex configuration or premium SKUs

• Distributed features (like CDC or read-only learners) were locked behind paywalls

With Dgraph v25, we’re changing that.

All enterprise features are now included by default, running on top of Hypermode’s serverless runtime. No keys. No carveouts. Just a fast, flexible graph engine that works how modern teams build.

What's new in Dgraph v25

Dgraph v25 is the result of focused development, community feedback, and rigorous performance testing. Let’s walk through the most significant changes.

Multi-tenant graphs with namespaces

Managing multi-customer apps within a single Dgraph cluster is now easier than ever. With namespaces, you can:

• Logically scope different customers
• Isolate data and permissions across different tenants while maintaining high performance
• Constrain vector searches within a single namespace
• Enhance scalability and lower operational complexity

Namespaces is now the standard approach for isolated data domains, replacing what is currently known as multi-tenancy. Each namespace is a fully scoped graph environment with its own schema and data that can be provisioned and queried independently. This shift makes Dgraph easier to use when operating across business domains or in multi-customer environments.

Supercharged AI coding with Model Context Protocol (MCP) servers

Make your coding assistant more powerful with access to your Dgraph cluster. The new MCP Servers allow you to:

• Incorporate your schema and data within Cursor, Windsurf, Claude Desktop, and more
• Choose between full access and read-only tools for enhanced control
• Ask your coding assistant to write queries and mutations
• Generate CRUD functions based on your schema

MCP servers bring your coding tools closer to your data, enabling smarter, context-aware code generation and faster development. By connecting your Dgraph cluster directly to your favorite assistants, you streamline onboarding, automate routine tasks, and get more relevant suggestions—right where you work.

Simplified development with v2 APIs

Dgraph v25 introduces the new v2 APIs helping you write less code and build faster. Available today in dgo, these APIs offer:

• Simplified client connections with support for Dgraph connection strings
• Unified functions for schema management, queries, mutations, and upserts
• Easy namespace management to create, rename, list, and drop namespaces

Designed for improved developer productivity, the v2 APIs make interacting with your Dgraph cluster smoother than ever. Adoption of the new APIs is optional and there are no plans to deprecate the v1 APIs. SDK updates for other languages are coming soon.

Enterprise-grade features now available for all

With Dgraph v25, powerful features previously reserved for enterprise customers are now available to everyone. These capabilities help you build secure, resilient, and compliant applications at any scale:

• Protect sensitive data with database encryption
• Track every request with comprehensive query logging
• Perform backups directly to S3-compatible storage
• Monitor and respond to data changes with Change Data Capture (CDC)
• Deploy read-only nodes to offload read-only workloads

These features, once exclusive to enterprise deployments, are now standard in Dgraph v25. Whether you’re building for regulated industries, need advanced security, or want to ensure business continuity, you can now leverage enterprise-grade reliability and control—no special license required. This release makes it easier than ever to adopt Dgraph for mission-critical workloads.

Powered by Hypermode infrastructure

All of this runs on Hypermode. That means:

• Fully serverless: spin up graphs in seconds, no cluster management
• Seamless scale: move to production with high-availability, no downtime required
• Complete AI workbench: use Model Router to add vector search or build agents on top of your graph, all from a unified experience

All new graphs created on Hypermode now run the preview of v25. Existing graphs are not impacted and will continue to run as configured.

What’s next

Beyond optimizing the getting started experience for the enterprise feature set, the path for Dgraph v25 to general availability includes several capabilities and upcoming improvements that we’re especially excited about:

• Unified import command: a new consolidated dgraph import command simplifies and speeds up importing data for new and running clusters.
• Native user auth: integrated authentication and authorization makes securing your cluster easier

The features in version 25 are part of our vision to make Dgraph the graph engine for agents and AI apps. Expect rapid iteration here as we gather feedback and move toward stable releases.

Try the preview

Dgraph v25 is available today as a public preview. You can:

• Spin up a local instance
• Deploy to Hypermode and test in a serverless runtime
• Explore new APIs and experimental features
• Join the community

We’re actively looking for feedback from teams of all sizes to help shape the final release.