Scheduled Stable Release November 17, 2026 (Currently on Beta channel)
Introduces a universal selector for custom highlights, ::highlight(*), which matches the highlight pseudo-elements of all custom highlights regardless of their name.
The * argument contributes zero specificity, so its declarations act as a base style that rules naming a specific highlight override through the cascade.
Introduce the flow-start/end keywords as value aliases for the existing flex-start/end keywords as resolved in the CSSWG. These are needed for alignment in the case of placement reversals.
flow-start in CSS Grid-Lanes would indicate the starting edge of the container at which placement begins, which is useful in the case where placement is reversed (fill-reverse, track-reverse, discussed in this CSSWG issue: [css-grid-3][masonry] item-flow row vs. column in masonry layouts · Issue #12803 · w3c/csswg-drafts). Because these are acting as value aliases, we expect flex-start/end to follow the logical directions in Grid Lanes, as well, which will expand the definition of these keywords from only applying special behavior in flexbox to applying to any layout mode whose logical start/end is different from its physical start/end.
For serialization, if either flex-start/end or flow-start/end is used, the returned serialized value should be flow-start/end.
This feature lets a single-axis scroll container (for example, one with overflow: auto clip) handle scroll snap area propagation and alignment separately for each axis.
Single-axis scroll containers are now supported (https://chromestatus.com/feature/5067363861004288). Without this feature, such a container captures a snap area on both axes, even the axis it can't scroll. The snap area then can't align on that axis.
With this feature, if a snap area sets alignment on both axes, the single-axis container handles only the axis it scrolls. The other axis goes up to the nearest ancestor scroll container that scrolls in that direction.
SVG documents loaded through <object> or <embed> are sized and positioned inside the element's content box according to the CSS object-fit and object-position properties, the same way <img> content is.
Previously Chrome ignored both properties for embedded SVG documents and always stretched the document viewport to the whole content box.
Chrome can now trigger a request for the browser to install a web app, given a manifest URL and optional manifest ID. The <install> element enables cross-origin web app installation without JavaScript and provides a better developer experience than handling beforeinstallprompt events. The element is launching in parallel with navigator.install(), the imperative entry point to the same installation capability.
Admins can control this in two ways:
navigator.install() and <install>.Starting in Chrome 158, the new Web Install API allows websites to prompt users to install a web app, given a manifest URL and optional manifest ID. The navigator.install() method enables cross-origin web app installation and provides a better developer experience than handling beforeinstallprompt events. This API is launching in parallel with the <install> element, a declarative entry point to the same installation capability.
Admins can control this in two ways:
navigator.install() and <install>. OrThe :playing, :paused, :seeking, :buffering, :stalled, :muted, and :volume-locked CSS pseudo-classes match <audio> and <video> elements based on their state.
This is one of the focus areas in https://wpt.fyi/interop-2026.
This feature adds an incremental rendering mechanism to OfflineAudioContext so that the audio graph data can be incrementally processed rather than allocating the whole audio buffer up front. By incrementally delivering the data using data chunks, there is no need to have high memory requirements for the user's machine when processing high amounts of data.
Gives embedders control over programmatic focus from embedded content via the focus-without-user-activation permissions policy. When the policy is denied for the initiating frame, programmatic focus changes (from element.focus(), autofocus, window.focus(), and the focusing steps for dialogs and popovers) are blocked unless that frame has transient user activation or focus is already within its subtree. This does not prevent dialogs or popovers from opening. User-initiated focus such as clicking or tabbing is never affected. Chromium's existing restriction on autofocus in cross-origin iframes is unchanged: the policy can further restrict autofocus, but allowing it does not enable autofocus in a cross-origin frame.
The policy can be set via a Permissions-Policy HTTP response header or the iframe allow attribute. Focus delegation is supported: a parent frame that has focus can programmatically pass focus to a child iframe, even if the child has the policy denied, and once a frame has focus it can move focus within its own subtree.
Prerenders cross-origin iframes with an opt-in response header.
Browsers will now prerender all cross-origin frames if the top-level frame's HTTP response includes the Supports-Loading-Mode: prerender-cross-origin-frames.
The circle() and ellipse() CSS basic-shape functions accept the closest-corner and farthest-corner radius keywords, in addition to the existing closest-side and farthest-side. These keywords resolve to the Euclidean distance from the shape center to the nearest or farthest corner of the reference box, matching the long-standing behavior of radial-gradient(). They work in clip-path, shape-outside, and offset-path, so the same shape syntax accepted by gradients now works for shapes.
CL: https://chromium-review.googlesource.com/c/chromium/src/+/7767079
Adds Source-Specific Multicast (SSM) support to the Direct Sockets API, allowing Isolated Web Apps to optionally specify a source address when joining multicast groups to receive UDP packets only from that source. Without a source address, joinGroup() uses Any-Source Multicast (ASM), which receives packets from any source sending to the group. SSM is essential to using multicast outside of private networks, simplifying packet routing, and filtering traffic at the network level by source address, which mitigates spoofing and denial-of-service attacks.
The WebHID API enables web applications to interact with Human Interface Devices (HIDs) over USB and Bluetooth beyond standard system input mappings (such as mice, keyboards, touchscreens, and standard gamepads). Web applications can request user permission to connect to HID devices, send and receive input, output, and feature reports, and inspect report descriptors.
WebHID shipped on desktop platforms (Windows, macOS, Linux, and ChromeOS) in Chrome 89. This launch brings the existing WebHID API to Chrome on Android in Chrome 158 (relying on new OS support in Android 17 / API level 37 and later) with no changes to the web-exposed API surface.
Popular uses today include customizing mechanical keyboards, keypads, and gaming mice, calibrating gamepads, and integrating hardware headset call controls in call centers.
WebView is not supported, as it does not have a device chooser UI or permissions UX affordances.
This feature can be controlled by the following enterprise policies:
Allows WebTransport applications to indicate how many incoming streams they expect to have open at once. Constructor hints configure the initial QUIC stream allowance, reducing the risk that a server that needs high stream concurrency is blocked by the initial stream limit.
Adds the congestionControl constructor option and read-only attribute to WebTransport. Applications can request "default", "throughput", or "low-latency" behavior, allowing the user agent to select suitable congestion-control behavior for each session.
WebTransport.draining is a promise that resolves when the server indicates that the WebTransport session will be gracefully retired. Applications can use this signal to establish a replacement session while existing streams and the current session remain usable.
Adds WebTransport.exportKeyingMaterial(), which allows an application to derive cryptographic keying material bound to an established WebTransport session.
The method accepts application provided binary label and context values and a requested output length and returns a Promise<Uint8Array>. Chromium supports label and context values up to 255 bytes and output lengths from 1 through 4096 bytes.
For WebTransport over HTTP/3, Chromium includes the WebTransport CONNECT stream ID in the TLS exporter context. This ensures that separate WebTransport sessions derive different keying material even when they share the same underlying HTTP/3 connection.
Adds the standardized WebTransportReceiveStream interface, a ReadableStream subclass used for incoming unidirectional streams and the readable side of bidirectional streams. Its getStats() method returns per-stream byte counters: bytesReceived reports bytes received from the network, while bytesRead reports bytes delivered to the application.
Adds the WebTransport.reliability instance attribute and static WebTransport.supportsReliableOnly attribute. These APIs indicate whether the user agent supports WebTransport over exclusively reliable connections and whether an established session supports unreliable transport such as datagrams.
In Chromium, reliability initially returns "pending" and changes to "supports-unreliable" after an HTTP/3 WebTransport connection is established. Since Chromium does not currently support HTTP/2 fallback, WebTransport.supportsReliableOnly returns false; any successfully established session uses HTTP/3 and supports unreliable datagrams.