Gel is an open-source graph-relational database, written in Python and released under Apache-2.0, that gives application developers a declarative object-type schema, the EdgeQL query language, a migrations system, client libraries and a command line tool in place of hand-written SQL and a separate ORM layer.

What it is

Gel is a database that takes the best parts of relational databases, graph databases and ORMs — a combination the project calls a graph-relational database. It sits in the Infrastructure & Operations / Databases category of this registry and is developed in the open on GitHub, where it carries Apache-2.0 licensing, roughly 14,170 stars and 450 forks. Its schema is written in ESDL and modelled the way application code models data: as object types containing properties connected by links, rather than tables joined by foreign keys. The README's own example is a type Person { required name: str; } linked from a type Movie { required title: str; multi actors: Person; }.

The problem it solves is the object-relational impedance mismatch. Tabular data modelling is described as a relic of an older age that is not compatible with modern languages, so Gel replaces both the relational schema layer and the ORM or mapper library that usually sits on top of it. Queries are written in EdgeQL, a ground-up redesign of SQL that returns rich structured objects instead of flat rows, which makes deeply fetching related objects painless and removes the need for JOINs. A query such as select Movie { title, actors: { name } } filter .title = "The Matrix" returns the movie and its actors as one object.

Key capabilities

• Schema declared in ESDL as object types with properties and links, replacing tables, foreign keys and a strict type system layered over tabular storage.
• EdgeQL, a composable query language in which one query can be used as an expression inside another, making subqueries and nested mutations straightforward.
• Comprehensive EdgeQL standard library, computed properties, polymorphic queries, with blocks and transactions.
• Support for everything expected from a database plus link properties, schema mixins and best-in-class JSON support.
• A migrations system for evolving schema over time.
• A suite of client libraries in different languages, plus a command line tool for creating instances, running migrations and opening an interactive EdgeQL shell.
• EdgeQL examples in the README cover reads (select Movie { ... } filter .title = "The Matrix") and nested writes (insert Movie { title := "...", actors := (select Person filter .name in {...}) }).

Who uses it and how

• Application teams that want their schema to be readable, writable and understandable as code, rather than a set of table definitions that drift from the application's own types.
• Projects replacing an ORM or mapper library with a database that handles the same job natively, keeping migrations, queries and the type system in one place.
• Teams following the 10-minute quickstart, which the README points to as the fastest route to a running instance, including a Next.js path.
• Groups that prefer not to operate the database themselves, using the managed Gel Cloud service described in the README.
• Contributors building Gel from source, following the contributing guide to compile it on a local machine and filing issues or discussions through GitHub or Discord.

Getting started

Install the CLI with curl --proto '=https' --tlsv1.2 -sSf https://geldata.com/sh | sh, or the PowerShell equivalent iwr https://geldata.com/ps1 -useb | iex on Windows, then run edgedb project init and edgedb to open an interactive EdgeQL shell. Gel Cloud is offered as the managed hosting option.

How it compares

This registry entry names no specific competing database or ORM products, so Gel stands alone here among graph-relational databases. Its own positioning is as a hybrid: it aims to combine the strengths of relational databases, graph databases and ORMs rather than occupying one of those categories. Teams comparing it should treat it as a replacement for both a relational database and the ORM layer written against it, not as an add-on to either.

When to use it — and when not to

A self-hoster must operate a database instance along with the CLI, the ESDL schema and the migrations system that Gel introduces in place of plain SQL files. Teams that need drop-in SQL compatibility, existing SQL tooling or a schema they are unwilling to redefine in object types should not pick it, because EdgeQL is a redesign of SQL rather than a dialect of it. The registry entry also carries a caution worth noting: 948 open issues and a README excerpt whose licence section is cut off mid-sentence, so prospective users should read the full documentation and the repository itself before committing.

type Person {
  required name: str;
}

type Movie {
  required title: str;
  multi actors: Person;
}

select Movie {
  title,
  actors: {
    name
  }
}
filter .title = "The Matrix"

insert Movie {
  title := "The Matrix Resurrections",
  actors := (
    select Person
    filter .name in {
      'Keanu Reeves',
      'Carrie-Anne Moss',
      'Laurence Fishburne'
    }
  )
}

$ curl --proto '=https' --tlsv1.2 -sSf https://geldata.com/sh | sh
$ edgedb project init
$ edgedb
edgedb> select "Hello world!"

PS> iwr https://geldata.com/ps1 -useb | iex