Edit

Microsoft Edge 159 web platform release notes (Dec. 10, 2026)

The following are the new web platform features and updates in Microsoft Edge 159, which releases on Dec. 10, 2026.

To stay up-to-date and get the latest web platform features, download a preview channel of Microsoft Edge (Beta, Dev, or Canary); go to Become a Microsoft Edge Insider.

Detailed contents:

Microsoft Edge DevTools

See What's new in Microsoft Edge DevTools.

WebView2

See Release notes for WebView2.

CSS features

The following new Cascading Style Sheets (CSS) features are included in Microsoft Edge.

AccentColor and AccentColorText follow accent-color

The AccentColor and AccentColorText system colors now reflect the element's effective accent-color value, instead of only the accent colors that are provided by the platform.

Use the AccentColor and AccentColorText system colors to reuse, in your own styles, the same colors that accent-color applies to form controls and other accented UI.

For example, the following CSS gives a custom element the same accent color as the checkbox that's next to it:

.option {
  accent-color: rebeccapurple;
}

.option .custom-marker {
  background: AccentColor;
  color: AccentColorText;
}

See also:

Computed background values keep their layer count

The computed values of the background-* longhand properties, such as background-repeat and background-size, now keep the number of comma-separated values that you specified. Previously, these computed values were repeated or truncated to match the number of layers in background-image.

Repeating and truncating still happen when the browser renders the layers, because they're part of the used value, not the computed value. As a result, this change only affects what you read back, such as by using getComputedStyle().

For example, in the following CSS, background-repeat has two values, while background-image has three layers. The computed value of background-repeat still has the two values that you specified.

.hero {
  background-image: url(a.png), url(b.png), url(c.png);
  background-repeat: no-repeat, repeat;
}

See also:

Scroll snap in single-axis scroll containers

Scroll snap now works separately for each axis in single-axis scroll containers, such as an element that has overflow: auto clip.

Previously, a single-axis scroll container captured snap areas on both axes, even on the axis that it can't scroll. A snap area then couldn't align on that axis.

Now, when a snap area sets alignment on both axes, a single-axis scroll container only handles the axis that it scrolls. The other axis goes up to the nearest ancestor scroll container that scrolls in that direction.

See also:

CSSImportRule.styleSheet can be null

The styleSheet property of the CSSImportRule interface now returns null when an @import rule has no associated stylesheet, such as when the supports() condition of the rule is false.

Use the CSSImportRule.styleSheet property to check whether a stylesheet is associated with an @import rule, before you access its properties:

const importRule = document.styleSheets[0].cssRules[0];
if (importRule.styleSheet) {
  console.log(importRule.styleSheet.cssRules.length);
}

See also:

Web APIs

The following new Web API features are included in Microsoft Edge.

Sanitizer API: javascriptURLs option

The Sanitizer API now provides the javascriptURLs configuration option and the setJavascriptURLs() method. The javascriptURLs configuration option and the setJavascriptURLs() method control whether javascript: URLs that are used for navigation (such as in the href attribute of an <a> element) are removed when sanitizing HTML.

Previously, whether javascript: URLs were removed depended on the method that you used:

  • The safe methods, such as setHTML(), always removed javascript: URLs.
  • The unsafe methods, such as setHTMLUnsafe(), never removed javascript: URLs.

Now, removing javascript: URLs is an explicit part of the sanitizer configuration, which you can use with any method.

If you don't use the new option, the behavior is unchanged.

See also:

Prerendering cross-origin iframes

On a top-level page, to prerender all of the page's cross-origin iframes along with the page, use the following HTTP response header:

Supports-Loading-Mode: prerender-cross-origin-frames

By default (that is, without using that response header), when a page is prerendered, the browser delays loading its cross-origin iframes until the prerendered page is activated. When cross-origin iframes are critical to your application, this delay can negate many of the benefits of prerendering.

See also:

Media Session API: setMicrophoneActive() and setCameraActive() return promises

The navigator.mediaSession.setMicrophoneActive() and navigator.mediaSession.setCameraActive() methods now return a Promise that resolves with undefined, instead of directly returning undefined.

If the associated document isn't fully active, the promise is rejected with an InvalidStateError exception.

See also:

Service worker static routing: Distinct error for cancelled raced requests

When a service worker uses a static routing rule with the race-network-and-fetch-handler source, and its fetch event handler calls fetch(event.request), the browser reuses the raced network request instead of sending a second network request.

If the raced network request is cancelled before the response headers arrive, the promise that's returned by fetch(event.request) is now rejected with a TypeError that has a specific message: The service worker race-network-and-fetch-handler request was cancelled before 'fetch()' settled. Previously, the message was the generic Failed to fetch message.

For example, a raced network request gets cancelled in the following situations:

  • When your offline fallback timer resolves event.respondWith() first.
  • When the navigation is aborted.

Use the new message to distinguish between these expected cancellations versus network errors.

See also:

Web Speech API: Timestamps for SpeechRecognitionResult

The SpeechRecognitionResult interface now has the speechStartTime and speechEndTime properties. The speechStartTime and speechEndTime properties give the start and end times of the speech segment that was recognized, in seconds, relative to the start of the audio stream.

Previously, the Web Speech API didn't expose when a segment of speech occurred. Use the speechStartTime and speechEndTime properties to associate transcribed text with a media timeline, such as to generate synchronized subtitle cues.

See also:

WebTransport: Datagram readable stream type

The WebTransport() constructor now accepts the datagramsReadableType option.

By default, incoming datagrams are exposed through a default ReadableStream. If your application needs to read datagrams into buffers that you provide, by using a BYOB (bring your own buffer) reader, set datagramsReadableType to "bytes" to get a readable byte stream instead:

const transport = new WebTransport(url, { datagramsReadableType: "bytes" });

See also:

WebTransport: Datagram writable streams and prioritization

The createWritable() method of WebTransportDatagramDuplexStream now creates independent WebTransportDatagramsWritable streams, which you can use to send datagrams.

Each writable stream has mutable sendGroup and sendOrder properties, which express the relative priority of the queued data in the same send group. Microsoft Edge applies this ordering among datagram writable streams.

Microsoft Edge also reports the negotiated maximum size of outgoing datagrams, and discards writes that are larger than this size, as required by the WebTransport specification.

See also:

WebTransport: Send groups and send-stream prioritization

You can now assign outgoing WebTransport streams to send groups, and use the sendOrder property to prioritize streams within each group. Previously, you could send data by using WritableStream objects, but you couldn't control how the browser prioritized the streams.

Statistics about sending are also available for streams and for send groups. Transferring a WebTransportSendStream still works the same way as transferring a WritableStream.

See also:

WebTransport: WebTransportReceiveStream

The new WebTransportReceiveStream interface is a subclass of ReadableStream that WebTransport uses for incoming unidirectional streams, and for the readable side of bidirectional streams.

The getStats() method of the WebTransportReceiveStream interface returns the following byte counters for the stream:

  • bytesReceived - The number of bytes received from the network.
  • bytesRead - The number of bytes delivered to your application.

See also:

WebTransport: Congestion control

The WebTransport() constructor now accepts the congestionControl option, and WebTransport instances now have a read-only congestionControl property.

Set the congestionControl option to one of the following values, to tell the browser which behavior you prefer for a session:

  • "default" - Let the browser use its default congestion-control behavior.
  • "throughput" - Favor the amount of data that's sent over time.
  • "low-latency" - Favor a low delay in delivering data.

The browser uses your preference to select suitable congestion-control behavior for each session.

See also:

WebTransport: Keying material export

Use the new WebTransport.exportKeyingMaterial() method to derive cryptographic keying material that's bound to an established WebTransport session.

The exportKeyingMaterial() method takes binary label and context values that you provide, and the length of the output that you want. It returns a promise that resolves with a Uint8Array. The label and context values can be up to 255 bytes. The output length can be from 1 through 4096 bytes.

Different WebTransport sessions derive different keying material, even when they share the same underlying HTTP/3 connection.

See also:

WebTransport: WebTransportError alignment and serialization

WebTransportError now matches the current WebTransport specification:

  • It has the standard (message, options) constructor. The dictionary constructor that was already available still works, for compatibility.
  • It carries the full unsigned-long stream error code.
  • It's structured-clonable, so you can pass it to other contexts, such as by using postMessage(). The cloned error preserves its message, source, and nullable streamErrorCode.

See also:

Origin trials

The following are origin trials for new experimental APIs that are available in Microsoft Edge.

Origin trials let you try experimental APIs on your own live website for a limited time. To learn more about origin trials, see Use origin trials in Microsoft Edge.

For the full list of available origin trials, see Microsoft Edge Origin Trials.

Name Description Register
Autofill Event Detects when browser autofill updates form controls, for adapting custom UI and validation. Register
Speculative Load measurement performance.speculations exposes data about used and unused preloads, prefetches, and prerenders. This helps you measure the effectiveness of speculative loading strategies and optimize performance-versus-cost tradeoffs. Register
Prompt API sampling parameters The topK and temperature sampling parameters let you optimize model behavior per language-model session. Register
Digital Credentials API - Issuance Support Lets a site trigger the issuance of a credential, such as a driver's license or an academic degree, from an issuer server to the user's digital wallet application. Register
PrefetchAndPrerenderActivationBeacon Allows Microsoft Edge to reliably report an activation beacon when a prefetch or prerendered page is activated. Register
prerender_until_script Speculation Rules API action Prerenders a page, but switches to prefetching when a script executes. Register
JS Self-Profiling Markers Extends the JS Self-Profiling API by adding sample markers such as script, gc, style, layout, and paint. These markers identify the activity that the browser was doing at capture time. Register
Email Verification Protocol Enables websites to verify email ownership by using cryptographic tokens that are issued by email providers. This reduces reliance on email one-time passcodes (OTPs), and streamlines sign-up, sign-in, and recovery flows. Register
WebAssembly Custom Descriptors V2 Stores data that's associated with source-level types more efficiently, in custom descriptor objects. Register
Declarative Performance Observer API Uses HTTP response headers to prove a reliable browser-managed telemetry system that reports performance metrics for the site, from navigation initiation to page termination. Register
WebMCP Enables your site to register tools for an in-browser agent to complete tasks on behalf of a user. Register
WebRTC data channel: SCTP Negotiation Acceleration Protocol (SNAP) Speeds up the establishment of a WebRTC data channel, by embedding the Stream Control Transmission Protocol (SCTP) initialization parameters in the Session Description Protocol (SDP) offer and answer. This can save up to two network round trips when opening a data channel. Register
WebRTC Diagnostic Logging API Lets your web app start, finish, and cancel the collection of internal diagnostic logs for WebRTC operations. Register
Connection Allowlist Embedded Enforcement Lets you use the connectionallowlist attribute on an <iframe> element, to require that the embedded document restricts its outgoing network connections to a declared allowlist. If the embedded document doesn't deliver a matching Connection-Allowlist header, its response is blocked. Register

Note

Portions of this page are modifications based on work created and shared by Chromium.org and used according to terms described in the Creative Commons Attribution 4.0 International License.