AutoDisposeProvider<T> class Null safety

A provider that exposes a read-only value.

What is a provider

Providers are the most important components of Riverpod. In short, you can think of providers as an access point to a shared state.

Providers solve the following problems:

  • Providers have the flexibility of global variables, without their downsides.
    Providers can be accessed from anywhere, while ensuring testability and scalability.

  • Providers are safe to use.
    As opposed to most service-locator solutions, using a provider, it is not possible to read a value in an uninitialized state.
    If we can write the code to read a state, the code will execute properly. Even if the state is loaded asynchronously.

  • Providers allow easily and efficiently listening to a piece of state.
    They can be accessed in a single line of code, and offers many way to optimize your application.

Creating a provider

Providers come in many variants, but they all work the same way.

The most common usage is to declare them as global variables like so:

final myProvider = Provider((ref) {
  return MyValue();

NOTE Do not feel threatened by the fact that a provider is declared as a global. While providers are globals, the variable is fully immutable. This makes creating a provider no different from declaring a function or a class.

This snippet consist of three components:

  • final myProvider, the declaration of a variable.
    This variable is what we will use in the future to read the state of our provider. It should always be immutable.

  • Provider, the provider that we decided to use.
    Provider is the most basic of all providers. It exposes an object that never changes.
    We could replace Provider with other providers like StreamProvider or StateNotifierProvider, to change how the value is interacted with.

  • A function that creates the shared state.
    That function will always receive an object called ref as a parameter. This object allows us to read other providers or to perform some operations when the state of our provider will be destroyed.

The type of the object created by the function passed to a provider depends on the provider used.
For example, the function of a Provider can create any object. On the other hand, StreamProvider's callback will be expected to return a Stream.

NOTE: You can declare as many providers as you want, without limitations.
As opposed to when using package:provider, in Riverpod we can have two providers expose a state of the same "type":

final cityProvider = Provider((ref) => 'London');
final countryProvider = Provider((ref) => 'England');

The fact that both providers creates a String does not cause conflicts. We will be able to read both values independently from each other without issue.

WARNING For providers to work, you need to add ProviderScope at the root of your Flutter applications:

void main() {
  runApp(ProviderScope(child: MyApp()));

Combining providers

We've previously seen how to create a simple provider. But the reality is, in many situation a provider will want to read the state of another provider.

To do that, we can use the ref object passed to the callback of our provider, and use its watch method.

As an example, consider the following provider:

final cityProvider = Provider((ref) => 'London');

We can now create another provider that will consume our cityProvider:

final weatherProvider = FutureProvider((ref) async {
  // We use `` to watch another provider, and we pass it the provider
  // that we want to consume. Here: cityProvider
  final city =;

  // We can then use the result to do something based on the value of `cityProvider`.
  return fetchWeather(city: city);

That's it. We've created a provider that depends on another provider.

One interesting aspect of this code is, if city ever changes, this will automatically call fetchWeather again and update the UI accordingly.

Creating an object that depends on a lot of providers.

Sometimes, we may want to create an object that depends on a lot of providers like so:

final cityProvider = Provider((ref) => 'London');
final countryProvider = Provider((ref) => 'England');

final weatherProvider = Provider((ref) {
  final city =;
  final country =;

  return Location(city: city, country: country);

class Location {
  Location({required, required});

  final String city;
  final String country;

  String get label => '$city ($country)';

This can quickly become tedious.

In that situation, it may be reasonable to pass the ref variable to our object directly:

final cityProvider = Provider((ref) => 'London');
final countryProvider = Provider((ref) => 'England');

final weatherProvider = Provider((ref) {
  // Pass the `ref` object to our `Location` class.
  // `Location` will then be able to call `` to read the providers.
  return Location(ref);

class Location {

  final ProviderReference _ref;

  String get label {
    final city =;
    final country =;
    return '$city ($country)';

This avoids having to implement a constructor, which makes changes on the object easier.

This is fine as, as opposed to BuildContext from Flutter, that ref object is completely independent from Flutter/the UI.
As such the object can still be shared and tested.

Disposing the resources the state is destroyed

During the lifetime of an application, the state associated with a provider may get destroyed.
In this situation, we may want to perform a clean-up before the state destruction.

This is done by using the ref object that is passed to the callback of all providers.

That ref object expose an onDispose method, which can be used to listen to the state destruction even to perform some task.

The following example uses ref.onDispose to close a StreamController:

final example = StreamProvider.autoDispose((ref) {
  final streamController = StreamController<int>();

  ref.onDispose(() {
    // Closes the StreamController when the state of this provider is destroyed.


See also:

  • Provider.autoDispose, to automatically destroy the state of a provider when that provider is no-longer listened.
  •, to allow providers to create a value from external parameters.


AutoDisposeProvider(Create<T, AutoDisposeProviderReference> _create, {String? name})
A provider that exposes a read-only value. [...]


argument Object?
If this provider was created with the .family modifier, argument is variable used.
read-only, inherited
debugId String
A unique identifier for this provider, used by devtools to differentiate providers [...]
late, final, inherited
from Family<dynamic, dynamic, dynamic, ProviderReference, RootProvider>?
If this provider was created with the .family modifier, from is the .family instance.
read-only, inherited
hashCode int
The hash code for this object. [...]
@nonVirtual, read-only, inherited
name String?
A custom label for providers. [...]
final, inherited
runtimeType Type
A representation of the runtime type of the object.
read-only, inherited


create(covariant AutoDisposeProviderReference ref) → T
createElement() AutoDisposeProviderElement<T, T>
An internal method that defines how a provider behaves.
createState() → _AutoDisposeProviderState<T>
An internal method that creates the state of a provider.
noSuchMethod(Invocation invocation) → dynamic
Invoked when a non-existent method or property is accessed. [...]
overrideWithProvider(AutoDisposeProviderBase<Object?, T> provider) → ProviderOverride
Overrides the behavior of this provider with another provider. [...]
overrideWithValue(T value) Override
Overrides the behavior of a provider with a value. [...]
select<Selected>(Selected selector(T value)) ProviderListenable<Selected>
Partially listen to a provider. [...]
toString() String
A string representation of this object. [...]


operator ==(Object other) bool
The equality operator. [...]


family → const AutoDisposeProviderFamilyBuilder
A group of providers that builds their value from an external parameter. [...]
const AutoDisposeProviderFamilyBuilder()