Merge pull request #3778 from MicrosoftDocs/main

main to live Apr. 16, 2026

Author: mikehoffms

microsoft-edge/web-platform/release-notes/148.md

@@ -32,6 +32,7 @@ To stay up-to-date and get the latest web platform features, download a preview
   * [Manifest localization](#manifest-localization)
   * [Support for avar2 in OpenType font format](#support-for-avar2-in-opentype-font-format)
   * [Pointer event suppression on drag start](#pointer-event-suppression-on-drag-start)
+  * [Prompt API](#prompt-api)
   * [Reuse `no-store` images when same `src` is reassigned](#reuse-no-store-images-when-same-src-is-reassigned)
   * [Web Authentication Immediate UI mode](#web-authentication-immediate-ui-mode)
   * [WebGPU `linear_indexing` feature](#webgpu-linear_indexing-feature)
@@ -41,7 +42,7 @@ To stay up-to-date and get the latest web platform features, download a preview
   * [Writer API](#writer-api)
   * [Rewriter API](#rewriter-api)
   * [Proofreader API](#proofreader-api)
-  * [Prompt API](#prompt-api)
+  * [Prompt API](#prompt-api-1)
   * [WebGPU Compatibility Mode](#webgpu-compatibility-mode)
   * [Extended lifetime for Shared Workers](#extended-lifetime-for-shared-workers)
   * [SharedArrayBuffers in non-isolated pages on Desktop platforms](#sharedarraybuffers-in-non-isolated-pages-on-desktop-platforms)
@@ -291,35 +292,30 @@ See also:
 * [Pointer events](https://developer.mozilla.org/docs/Web/API/Pointer_events) at MDN.
 
 
-<!-- ------------------------------
-todo: verify if this is also shipping in Edge.
+<!-- ------------------------------ -->
 #### Prompt API
 
-The Prompt API provides direct access to a browser-provided on-device AI language model.
+The Prompt API provides direct access to a browser-provided on-device AI language model.  The API design offers fine-grained control for progressively enhancing sites with model interactions tailored to individualized use cases.
 
-The API design offers fine-grained control for progressively enhancing sites with model interactions tailored to individualized use cases.
+The Prompt API complements task-based language model APIs (such as the Summarizer API), and varied APIs and frameworks for generalized on-device inference using developer-supplied ML models.
 
-This complements task-based language model APIs (such as the Summarizer API) and varied APIs and frameworks for generalized on-device inference with developer-supplied ML models.
-
-The initial implementation supports text, image, and audio inputs, as well as response constraints that ensure generated text conforms with predefined regex and JSON schema formats.
-
-Use for:
+The initial implementation supports:
+* Text inputs.
+* Image inputs.
+* Audio inputs.
+* Response constraints that ensure that the generated text conforms with predefined regex and JSON schema formats.
 
+Use the Prompt API for:
 * Generating image captions.
-
 * Performing visual searches.
-
 * Transcribing audio.
-
 * Classifying sound events.
-
 * Generating text by following specific instructions.
-
 * Extracting information or insights from multimodal source material.
 
 See also:
-* [Prompt API](https://developer.chrome.com/docs/extensions/ai/prompt-api) on developer.chrome.com.
--->
+* [Prompt a built-in language model with the Prompt API](../prompt-api.md)
+* [Origin trials > Prompt API](#prompt-api-1), below.
 
 
 <!-- ------------------------------ -->
@@ -410,9 +406,7 @@ The Writer API can be used to generate text from a writing task prompt, backed b
 For example, use this API to:
 
 * Generate a textual explanation of structured data.
-
 * Compose a post about a product, based on reviews or a product description.
-
 * Generate a biography based on background information.
 
 See also [Summarize, write, and rewrite text with the Writing Assistance APIs](../writing-assistance-apis.md).
@@ -432,9 +426,7 @@ The Rewriter API transforms and rephrases input text in requested ways, backed b
 Use this API to remove redundancies within input text, in order to:
 
 * Fit the text into a word limit.
-
 * Rephrase the text to suit the intended audience.
-
 * Be more constructive if the text uses inadequate language.
 
 See also [Summarize, write, and rewrite text with the Writing Assistance APIs](../writing-assistance-apis.md).
@@ -472,16 +464,16 @@ The Prompt API is an experimental web API.
 Use the Prompt API to:
 
 * Generate and analyze text.
-
 * Create application logic based on user input.
-
 * Discover innovative ways to integrate prompt-engineering capabilities into your web app.
 
+See also:
+* [Prompt API](#prompt-api), above.
+
 Information about this origin trial:
 * [Explainer](https://github.com/webmachinelearning/prompt-api/blob/main/README.md)
 * [Feedback](https://github.com/webmachinelearning/prompt-api/issues)
 * [Register](https://developer.microsoft.com/microsoft-edge/origin-trials/trials/b7d35247-b855-4b08-b237-89e7a5056117)
-* [Register](https://developer.microsoft.com/microsoft-edge/origin-trials/trials/b7d35247-b855-4b08-b237-89e7a5056117)
 
 
 <!-- ------------------------------ -->
@@ -510,9 +502,7 @@ Set the `extendedLifetime` option when creating a shared worker, to keep the ser
 Such async operations include:
 
 * IndexedDB writes.
-
 * Encryption or compression.
-
 * Analytics requests.
 
 Information about this origin trial:
@@ -698,9 +688,7 @@ By default, WebAudio processes audio in fixed blocks of 128 sample-frames (a ren
 Use the WebAudio Configurable Render Quantum origin trial to specify a `renderSizeHint` option when creating an `AudioContext` or `OfflineAudioContext`, to request a particular render quantum size.
 
 * Pass an integer, to request a specific size.
-
 * Pass `"default"` (or omit the option), to use the default of 128 frames.
-
 * Pass `"hardware"`, to let the browser pick an optimal size for the current configuration.
 
 Information about this origin trial: