Comparison
|
Jun 4, 2026
Hygraph vs. Directus
Architecture, content modeling, admin UI, automation, AI, and pricing compared.
Matt Minor
Senior Director, Growth
Note from Directus
Yes, this is a Directus vs Hygraph page written by Directus. Yes, we know that's absurd. Every vendor publishes one of these and ranks themselves first. (Yes, that includes Hygraph.) So, if we have to play this game, we'd rather play it differently. What follows is the most honest comparison we could write.
Two backends, two opposite directions.
Hygraph and Directus both give you a backend with auto-generated APIs and an admin UI on top of structured content. They get there from opposite directions. Hygraph is a GraphQL-first SaaS: your content lives in their cloud, modeled in their web app, and served through a federated GraphQL API that can also stitch in remote sources (Shopify, S3, your REST APIs). Directus is a self-hosted (or managed) platform that sits on top of any SQL database you bring or spin up.
Pick Hygraph if you want a hosted GraphQL-first content platform with content federation across multiple sources, a CDN, and a workflow-heavy editorial product. Pick Directus if you want to own your data in a real SQL database, self-host (or use Directus Cloud), pay predictable infrastructure costs instead of operations-and-traffic pricing, and get native dashboards, visual automation, AI Assistant, MCP server, and realtime APIs out of the box.
The feature-by-feature view.
A few rows do most of the work. Hosting model (self-host vs SaaS-only) is the architectural fork. Where data lives (your database vs their cloud) is the data-sovereignty question. GraphQL-first + content federation is Hygraph's distinct technical bet. Native AI, MCP, Flows, Insights, realtime are capabilities Directus ships natively that Hygraph either ships in narrower form or doesn't ship at all.
Dimension | Directus | Hygraph |
|---|---|---|
License | Source-available (MSCL) | Proprietary (closed source) |
Hosting | Self-host (Docker, Node) or Directus Cloud | SaaS only, vendor cloud |
Schema Philosophy | Reads your existing SQL schema | Schema defined in Hygraph's web app or via Management API |
Supported Databases | Postgres, MySQL, MariaDB, MSSQL, SQLite, Oracle, CockroachDB | N/A (no database access) |
API Output | REST, GraphQL, WebSockets, GraphQL Subscriptions | GraphQL Content, GraphQL Management, GraphQL Mutations |
Admin UI | The Studio (Vue 3) | Hygraph Studio (React) |
Primary API shape | REST-and-GraphQL parity | GraphQL-first (REST is a thin layer) |
Modeling Primitives | Collections, fields, relationships, M2A, translations | Models, components, relations, locales, workflow stages |
Where Data Lives | Your SQL database | Hygraph's cloud, vendor-controlled |
Native Realtime APIs | WebSockets, GraphQL Subscriptions | No (webhooks only) |
Native AI Assistant | Yes, in the Studio | AI Actions (authoring-focused) |
Native MCP server | Yes | No |
Visual Automation | Flows | No (webhooks + Workflow stages) |
Native Dashboards | Insights | No |
Asset CDN | Bring your own | Built-in CDN with image transforms |
Localization | Translations on any field, all tiers | Locales as first-class; tier-gated |
Pricing model | Per-environment / per-seat or self-hosted free | Per-project + operations / traffic / records quotas |
The main architectural difference.
Source-available vs. closed SaaS.
Hygraph is proprietary, closed-source software. You don't read its source, fork it, or run it. You consume the SaaS.
Directus is distributed under a source-available license. (See directus.com/license for the live terms.) The source code is fully readable and modifiable, and free to self-host for the vast majority of use cases. It is not open source in the OSI sense, and we don't claim it is.
Practically: you can read Directus's source, you cannot read Hygraph's. You can self-host Directus, including air-gapped or on-prem; you cannot self-host Hygraph. You can fork Directus; you can't fork Hygraph. You can run Directus indefinitely without paying anyone; Hygraph requires an active subscription past the free tier.
For teams who say they want "open source" but really mean "no vendor lock-in," Directus's source-available, self-hostable model is materially closer than a closed-source SaaS.
Content modeling.
Both tools model structured content well. The shape of the primitives differs.
Models in Hygraph are typed content schemas. You define them in the web app (or via Management API), set up fields and validation, and Hygraph generates GraphQL types automatically. Components are reusable groups of fields. Relations connect models. Workflow stages (Draft, Published, custom) are first-class.
Collections in Directus are real database tables. Fields are columns. Relationships are foreign keys. Many-to-Any (M2A) lets a single field reference items from any number of unrelated collections, useful for flexible page composition, polymorphic relationships, and arbitrary cross-collection joins. Because the model is the database, the same data is trivially queryable by any system that speaks SQL.
Workflow. Hygraph's built-in stages are a real strength for editorial teams who need explicit publishing pipelines. Directus handles this through statuses and roles plus Flows for routing logic. Hygraph's primitive is more opinionated and quicker for that specific use case; Directus's approach is more general but requires more setup.
Content federation. Hygraph's distinct angle: stitching remote sources into the GraphQL API. Declare a Remote Source (Shopify, REST endpoints), Hygraph proxies and stitches it into the schema, consumers query everything through one endpoint. Directus federates across multiple SQL databases - related but different shape (SQL, not arbitrary GraphQL/REST).
Localization. Hygraph treats locales as first-class on every model, clean for editorial teams running translated sites. Locale availability and counts are gated by pricing tier. Directus stores translations as related rows in a translations table that can hang off any field, available on every tier.
The admin experience.
Directus ships the Studio, built in Vue 3 and framed as a workspace, not as a content-publishing tool. The same UI that lets a content editor edit a blog post lets an operations person build a dashboard, configure a workflow, manage permissions, or chat with the built-in AI Assistant.
Hygraph ships Hygraph Studio, built in React. It's shaped primarily around editorial work: content authoring, scheduled publishing, stage-based workflows, multi-locale workflows, and developer ergonomics for managing schemas.
If your team mixes content people, ops people, analysts, AI agents, and developers who all need to work on the same governed data layer, the Directus Studio is built for that mix and Hygraph's admin isn't.
Automation and workflows.
Directus has Flows, a visual automation builder. Drag together triggers, conditions, and operations to build pipelines that run on data changes, schedules, or webhooks. Non-developers can build moderately complex automations. Developers can drop into custom JavaScript at any step.
Hygraph does not ship a visual automation builder. Automation comes from two places: webhooks (fire on content events to your own service) and Workflows (the editorial stage system - Draft, In Review, Published, with role-based gates between stages).
Realtime, federation, delivery.
Three capabilities that show up differently and matter at evaluation time.
Realtime APIs. Directus ships native WebSockets and GraphQL Subscriptions in the core. Subscribe to changes on any collection and push live updates to clients without adding infrastructure. Hygraph does not ship native realtime.
Federation. Hygraph's content federation stitches remote GraphQL or REST sources into one Content API. Directus federates across multiple SQL databases.
Content delivery. Hygraph ships a global CDN with on-the-fly image transformations. Directus is built to be unopinionated. You bring your own caching layer (CloudFront, Cloudflare, Fastly) and your own asset CDN. Directus ships asset transformations and signed URLs natively.
AI.
The AI Assistant lives within your Studio and can query/interact with your data.
Directus ships with a native AI Assistant inside the Studio. It's conversational and it takes action. It can create content, translate fields, summarize records, route items for review, and operate against your data with the same access policies as a human user. It's not a separate "AI mode," it's part of the same workspace.
Directus also runs a native MCP server. External AI tools (Claude Desktop, Cursor, ChatGPT, your own agent) can connect and work with the same data using the same access policies.
Hygraph has been adding AI features (AI Actions, content suggestions, asset alt-text, auto-translation), most surfaced inside Hygraph Studio. There's no native MCP server.
Developer experience.
npx
directus-template-cli@latest init
Directus
Both have CLIs (Hygraph's is lighter and centered on the Management API). Both have TypeScript SDKs. Hygraph's GraphQL documentation covers the API surface thoroughly.
GraphQL-first is Hygraph's distinct DX bet. The API surface is GraphQL through and through, with content federation extending the same endpoint across remote sources.
Directus offers REST and GraphQL parity (both are first-class) plus WebSockets and GraphQL Subscriptions for realtime, and the SDK consumes any of these. More API surfaces, less specifically GraphQL-tuned.
Schema as code. Hygraph supports it via the Management API (script schema changes, version in your repo). Directus supports it via Schema Snapshots (snapshot the database schema, apply to other environments). Both work; the mental model differs.
What it actually costs.
Self-hosting (Directus only): Free. You bring infrastructure, do operations, take on backups and upgrades. Hygraph has no self-host option at any tier.
Managed cloud: Directus Cloud has starter, team, and growth-style tiers running from $99/month to a few hundred dollars per month. Hygraph's paid tiers run on a per-project base plus metered usage.
The thing that varies most: Hygraph's pricing is per-project and metered on operations, traffic, and records. Teams routinely under-estimate Hygraph's total cost at scale because the metered line items grow with traffic and content volume, not with users. Model the total bill against your real expected traffic, record count, and locale count at your one-year scale, not the headline tier.
Enterprise: Both have enterprise tiers with SSO, audit logs, dedicated support, and SLAs. Both will quote you.
Neither is perfect.
Three areas where Hygraph is weaker | Three areas where Directus is weaker |
|---|---|
SaaS-only. No self-host, no on-prem, no air-gapped, no VPC deployment. Hard regulatory or sovereignty requirements rule it out. | No fully-zero-ops SaaS. Directus Cloud reduces ops, but Hygraph is true SaaS where you never think about infrastructure. |
No real database access. Your data lives in their cloud, queryable only through their GraphQL API. No SQL, no warehouse-native joins, no analytics jobs against the raw store, no sibling services sharing the same tables. | Hygraph's GraphQL-first DX is more tuned for GraphQL-native front-ends. Directus's GraphQL is solid but it's one API surface among several. |
Migration off Hygraph is significant work. Your data is in their content model, accessed only through their GraphQL API, and exporting at scale takes engineering effort. | Smaller marketplace of pre-built integrations than the larger SaaS CMSes. |
Who each is (usually) best for.
Directus is usually best for | Hygraph is usually best for |
|---|---|
Teams who want their content data to live in a SQL database they own, queryable by any service that speaks SQL | Teams that explicitly want zero-ops SaaS, are happy with a global CDN baked in, and don't want to think about infrastructure |
Teams with hard requirements around self-hosting, on-premise, air-gapped, EU data sovereignty, or regulated-industry compliance | Front-ends consuming content from many sources (CMS + Shopify + REST APIs + assets) that want one federated GraphQL endpoint |
Teams where content, operations, analytics, and AI agents all need to work on the same governed data layer | GraphQL-native engineering teams running Apollo, urql, or Relay, where API shape and tooling alignment matter |
Teams where non-developers own real workflows: dashboards (Insights), visual automations (Flows), AI-assisted operations | Teams running multi-locale content who want locales as a first-class primitive, and are on a tier that includes their locale counts |
Migration notes.
From Directus
Doable and well-trodden. The standard path: export your Hygraph content via the Management API or a GraphQL export script, transform the JSON into a relational schema (models become collections, components become repeating groups or related collections, relations become foreign keys, locales become translation rows), and import into your SQL database. UI Extensions don't transfer; rebuild as Directus extensions. Webhooks reattach to Flow triggers. Federation against remote sources needs to be re-architected. Directus federates across SQL, not across arbitrary GraphQL endpoints, so consumers depending on the unified Hygraph endpoint would call those sources directly or through a different layer.
From Hygraph
Harder. Hygraph's content model is more opinionated than a SQL schema, and any database features you've used (custom Postgres types, advanced indexes, M2A, multi-database joins, materialized views, raw SQL access) don't exist in Hygraph. You'd remodel your schema as Hygraph models and migrate data through the Management API. Anything outside Hygraph's content model (analytics queries, sibling services touching the database, scheduled jobs) needs to be re-architected to live somewhere else.
Common questions.
Is Directus better than Hygraph?
Neither is better in a general sense. Hygraph is a GraphQL-first SaaS for teams that want hosted content infrastructure with a federated API. Directus is a database-first backend for teams that want to own their data in a SQL database. The right choice depends on which model better matches your infrastructure, data ownership, and deployment requirements.
Can Hygraph be self-hosted?
No. Hygraph is SaaS-only at every tier. Your content lives in Hygraph’s cloud and is accessible through their GraphQL API.
What is content federation in Hygraph?
Content federation is Hygraph’s ability to connect remote sources such as Shopify, REST APIs, S3, and other services into a single GraphQL endpoint. Front ends can query one API and get content from multiple backends in one response. Directus federates across multiple SQL databases.
Does Directus have native realtime APIs that Hygraph does not?
Yes. Directus ships native WebSockets and GraphQL Subscriptions in core. You can subscribe to changes on any collection and push live updates to clients without extra infrastructure. Hygraph does not ship native realtime APIs. Live updates with Hygraph require webhooks and a separate realtime layer.
Does Directus have an MCP server?
Yes. Directus ships a native MCP server. External AI tools such as Claude, Cursor, ChatGPT, and custom agents can connect to Directus and operate under the same role-based permissions as human users. Hygraph does not ship a native MCP server. The free tier is usable for small projects. Growth-tier seats run in the mid-double digits per user per month, with overages on top. Teams routinely underestimate total cost at scale because the metered line items grow with traffic and content volume, not just headcount. Model your one-year bill before committing.
Can I migrate from Hygraph to Directus?
Yes. Export your Hygraph content through the Management API or a GraphQL export script. Then transform the JSON into a relational schema: models become collections relations become foreign keys locales become translation rows assets need to be mapped into your target file storage approach
