Hasura

This Hasura review covers what Hasura is used for, its key features, Hasura pricing for DDN and the older Cloud plans, how Hasura v2 differs from Hasura DDN, how it compares with Apollo, and where it falls short. Hasura connects to a database such as PostgreSQL, reads the tables and relationships, and serves a GraphQL API with queries, mutations, filtering, pagination and real-time subscriptions. Out of the box you do not write a GraphQL server or resolvers for standard data access. Access rules, custom business logic and events are configured on top. Hasura started as the free Hasura GraphQL Engine and now also sells Hasura DDN (Data Delivery Network), a hosted "supergraph" built on a rewritten engine.

What is Hasura used for?

Hasura is used to give frontend, mobile and backend teams a single data API without building one by hand. Typical uses:

• A backend for web and mobile apps on PostgreSQL, with role-based access for each user type.
• A unified GraphQL endpoint over several data sources: databases, existing REST APIs and other GraphQL services, so clients make one set of API calls instead of many.
• Real-time features such as live dashboards and chat through GraphQL subscriptions.
• Internal tools and admin panels that query the Hasura API from tools such as Retool or Appsmith.
• Replacing hand-written CRUD endpoints, which cuts development time on data-heavy modern applications.

Hasura is a developer tool, not a no-code app builder. You still design the database schema, write the permission rules and host or pay for the API. Low-code teams use it as the backend behind a front end they build elsewhere.

Key features

• Instant GraphQL APIs: the engine generates a GraphQL schema from your tables, views and relationships, including filtering, sorting, aggregations and pagination.
• REST APIs from GraphQL: turn any saved GraphQL query into a REST endpoint for clients that do not speak GraphQL.
• Authorization: row- and column-level permission rules per role, evaluated in the engine and driven by claims in your JWT or webhook. Security is where Hasura saves the most custom code.
• Custom business logic: Actions call your own HTTP handlers for logic that is not simple CRUD, and Remote Schemas merge other GraphQL servers into one endpoint.
• Event and cron triggers: call a webhook when a row is inserted, updated or deleted, or on a schedule.
• Data connectors: PostgreSQL and Postgres-compatible databases, plus MS SQL, BigQuery, Snowflake, MongoDB, MySQL, Oracle and other data sources (Hasura Cloud plans). On Hasura DDN, data connectors are separate services, so you can connect new data sources or write your own connector in TypeScript, Python or Go.
• Hasura console, CLI and migrations: the Hasura console lets you explore data, set permissions and test queries in the browser, and the CLI keeps database migrations and configuration as metadata files in Git, so every environment can be set up the same way.
• Performance and caching: the engine compiles a GraphQL query into a single SQL query where it can, and the paid Cloud plans add query caching and read replicas.

Hasura pricing

Hasura sells two hosted products, and the community edition of the engine is free to self-host. The prices below come from the Hasura pricing page and the Hasura Cloud plans docs as of September 2026.

Hasura DDN (the current platform):

• DDN Free: $0, unlimited models, one supergraph developer, 15 minutes of observability retention.
• DDN Base: $5 per active model per month, unlimited developers, 30 days of observability.
• DDN Advanced: $30 per active model per month, adds federated collaboration, multi-repo CI/CD and team-based access control.
• Private DDN: dedicated VPC or your own infrastructure on Base and Advanced. Enterprise support is quoted by sales.

An active model is a model (roughly a table or collection) that is queried more than 1,000 times a month (DDN pricing docs). API requests and concurrent users are not limited. Hosting for third-party and code connectors costs extra: $0.075 per vCPU-hour and $0.0075 per GiB-hour.

Hasura Cloud v2 (the classic hosted engine):

• Cloud Free: up to 3 projects, 3 million API requests and 100 MB of data passthrough a month, no credit card.
• Cloud Professional: $1.50 per active project hour for Postgres, $3.00 for other databases, plus $0.13 per GB of data passthrough.
• Cloud Enterprise: annual contract with dedicated infrastructure and 24/7 support.

The cost surprise to check is the billing metric. On DDN every model that crosses 1,000 requests a month counts, so a schema with many small tables can cost more than its traffic suggests. On Cloud v2 you pay for every hour a project is active plus data transfer. Self-hosting the free engine has no license fee, but you pay for the servers and run upgrades yourself. Check which features you need on each plan before you create a production project.

Performance, security and scalable setup

Performance is the main reason teams pick Hasura over a hand-written GraphQL server. The engine turns a nested GraphQL query into one SQL query instead of many resolver calls, which keeps complex queries fast on large tables. For a scalable system you run several stateless engine instances behind a load balancer, add read replicas and use query caching on the paid Cloud plans. Performance optimization mostly happens in the database: indexes, views and well-designed relationships.

Security is handled in the engine. Every request carries a role, and the permission rules decide which rows and columns that role can read or change, so you do not repeat access checks in each client. Admin access is protected by a secret, and production setups should disable the console and introspection for public roles. Hasura's documentation covers these security steps in detail, and the GraphQL community around it has published many example projects. Using Hasura well means treating performance and security as part of the setup, not something you add later.

Hasura v2 vs Hasura DDN

Hasura v2 is the original engine, written in Haskell and released under the Apache 2.0 license. It runs as a single server that you self-host with Docker or run on Hasura Cloud. Hasura DDN (Data Delivery Network) is built on engine v3, rewritten in Rust. It builds a supergraph from subgraphs that different teams own, uses data connectors for each source, and is configured as metadata files through the DDN CLI. The core v3 engine and the data connectors are Apache-licensed, while the DDN CLI and console are proprietary (Hasura on GitHub).

In June 2025 the company launched PromptQL, an AI data-access product, and said that Hasura GraphQL Engine and Hasura DDN "will continue to be hosted and maintained" (Hasura blog). New projects that want the managed platform create a supergraph on Hasura DDN. Teams that want a single GraphQL server they fully control still use v2. Hasura DDN is where new features land, so it is the safer choice for the future; v2 is stable and still maintained.

Hasura vs Apollo and other GraphQL solutions

Apollo and Hasura solve different parts of the GraphQL stack. With Apollo Server you write the GraphQL schema and resolvers yourself in code, and Apollo GraphOS adds a schema registry, Apollo Federation for a federated graph and managed routing. Apollo Client is a frontend library that works with any GraphQL API, including Hasura. Hasura generates the schema and resolvers from your data, so a standard data API takes minutes instead of weeks, and complex queries become one SQL query. Choose Apollo when most of your API is custom logic over many services. Choose Hasura when most of it is data access over databases. Other GraphQL solutions such as PostGraphile, StepZen and Grafbase sit between the two. Hasura's own comparison is written by the vendor and argues for Hasura.

Where Hasura shines — and where it falls short

Strengths:

• A production GraphQL API over an existing database in minutes, with no resolver code.
• Fine-grained, declarative authorization per role, row and column.
• Real-time subscriptions, event triggers and REST endpoints included.
• A free engine you can self-host, with migrations and metadata kept in Git.

Trade-offs:

• Two platforms with different engines and pricing models make the choice confusing.
• Custom business logic lives outside Hasura, in Actions or separate services.
• DDN's per-active-model pricing is hard to forecast for large schemas.
• The data model and permissions still need a developer; it is not a no-code tool.
• Enterprise features such as caching limits, dedicated infrastructure and SLAs sit on paid or Enterprise plans.

How Hasura compares

• Supabase: a hosted Postgres backend with auth, storage and auto-generated APIs in one product.
• NocoDB and Baserow: spreadsheet-style interfaces and APIs on a database, aimed at non-developers.
• Airtable: a no-code database with its own API, not built for large production apps.
• Retool and Appsmith: internal-tool builders that often sit on top of a Hasura API rather than replace it.
• n8n: workflow automation that can call Hasura's GraphQL or event webhooks.

Is Hasura worth it?

Hasura is worth it for developer teams that want a secure, fast GraphQL or REST API over PostgreSQL and other databases without writing and maintaining resolver code. The open source engine is free, and the hosted free tiers are enough to evaluate it. It is less of a fit when most of your API is custom business logic, when you want a no-code backend with a visual interface, or when you need predictable pricing for a very large schema on DDN. If you want help designing a Hasura backend, moving from v2 to DDN or building the front end on top, backend developers and low-code agencies on LowCodeDevs can take it on.

Sources

• Hasura pricing: DDN Free, Base, Advanced and Private DDN.
• Hasura DDN plans and pricing docs: the active model definition.
• Hasura Cloud plans: Cloud v2 Free, Professional and Enterprise, database rates and data passthrough.
• From GraphQL to PromptQL, Hasura blog: PromptQL launch and continued maintenance of HGE and DDN.
• Hasura GraphQL Engine on GitHub: v2 and v3 licenses.
• Hasura vs Apollo, Hasura blog: written by the vendor.
• Up and Running with Hasura, Medium: hands-on setup with Docker, the CLI and migrations.
• Grafbase vs Hasura, Grafbase: written by a competitor. Its claims that Hasura does not support federation, can only be configured in the UI and has limited self-hosting conflict with Hasura's docs: DDN is built for federated supergraphs, both engines are configured as metadata files through the CLI, and the v2 engine is open source and self-hostable.
• Hasura raises $100 million Series C, TechCrunch: funding and valuation.