Summarizer

Experimental: This is an experimental technology
Check the Browser compatibility table carefully before using this in production.

The Summarizer interface of the Summarizer API contains all the functionality for this API, including checking AI model availability, creating a new Summarizer instance, using it to generate a new summary, and more.

Instance properties

expectedContextLanguages Read only Experimental: The languages the context strings should be written in.
expectedInputLanguages Read only Experimental: The languages the Summarizer should support.
format Read only Experimental: The text format summaries will be returned in.
inputQuota Read only Experimental: The input quota available to the browser for generating summaries.
length Read only Experimental: The relative length of the generated summaries.
outputLanguage Read only Experimental: The language the summary should be generated in.
sharedContext Read only Experimental: A text string describing the context the pieces of text to summarize are being used in, which helps the Summarizer generate more suitable summaries.
type Read only Experimental: The type of summary that will generated by the Summarizer.

Static methods

availability() Experimental: Returns an enumerated value that indicates whether the browser AI model supports a given Summarizer configuration.
create() Experimental: Creates a new Summarizer instance from which to generate summaries.

Instance methods

destroy() Experimental: Destroys the Summarizer instance it is called on.
measureInputUsage() Experimental: Reports how much input quota would be used by a summarize operation for a given text input.
summarize() Experimental: Generates a new summary string.
summarizeStreaming() Experimental: Generates a new summary as a ReadableStream.

Examples

See Using the Summarizer API for a complete example.

Creating a `Summarizer` instance

const summarizer = await Summarizer.create({
  sharedContext:
    "A general summary to help a user decide if the text is worth reading",
  type: "tl;dr",
  length: "short",
  format: "markdown",
  expectedInputLanguages: ["en-US"],
  outputLanguage: "en-US",
});

Generating a summary

const summary = await summarizer.summarize(myTextString);
console.log(summary);

Generating a summary stream

const stream = summarizer.summarizeStreaming(myTextString);
let summary = "";

for await (const chunk of stream) {
  summary += chunk;
}

console.log("Stream complete");
summaryOutput.textContent = summary;

Specifications

Specification
Writing Assistance APIs # summarizer