Matter Reached 1.6 in June. Amazon and Google Are Still on 1.2: What ‘Matter-Ready’ Buys You

The specification and the ecosystems that implement it have come apart. A Matter logo tells a buyer which document a device was certified against, not what will work in their house.

Matter 1.6 was published on 17 June 2026, specification and software development kit together. The two platforms most American buyers actually own, Amazon’s and Google’s, are reported to be running Matter 1.2 to 1.3. That gap of three or four releases is not a footnote about roadmaps; it is the reason a certified device can arrive, pair correctly, and still refuse to do the thing the box implied. This guide is about the size of that gap and how to shop around it, not about which logo looks most reassuring.

Key facts

• Matter 1.6 released 17 June 2026, with NFC commissioning, Joint Fabric and thermostat suggestions
• matter-smarthome.de reports Amazon and Google still on Matter 1.2 to 1.3
• Apple Home supports selective Matter 1.4 features, including robot vacuums, while lacking most of 1.4
• Google Home has not released basic Matter 1.0 generic switches to users
• Thread 1.4 has been the sole border-router certification standard since 1 January 2026

What arrived on 17 June, and who can actually use it

The Connectivity Standards Alliance shipped three headline capabilities in 1.6. NFC-based commissioning is bi-directional and works before a device is fully powered on, so a bulb can be set up before it is screwed into the fixture and an in-wall switch before the mains is connected. Joint Fabric lets several user-authorised controllers co-administer one shared network through a central datastore, replacing the older Enhanced Multi-Admin approach. Thermostat Suggestions changes ecosystems from issuing commands to sending time-bound suggestions tied to presets, which the thermostat then weighs against user preference, so competing automations stop fighting each other.

Underneath that sit standardised device capability and limits reporting, event history for security sensors, an unmounted-state indication for smoke and carbon-monoxide alarms, and partitioned certificate revocation lists.

Every one of those features needs the controller in your house to implement it. Until Amazon, Google, Apple or Samsung ship the corresponding support, a device certified to 1.6 behaves in your home exactly like a device certified to whatever version your platform is on.

Amazon and Google are two versions behind the spec

The 2026 status review published by matter-smarthome.de puts numbers on this. Amazon and Google are still on 1.2 to 1.3. Apple Home supports selective 1.4 features, robot vacuums among them, while lacking most of 1.4. Google Home, four years into the standard, has not released basic Matter 1.0 generic switches to users.

That last detail deserves a moment. Generic switches were in the very first version of the specification. A buyer holding a Matter-certified button in a Google household in 2026 may find the platform has still not shipped the feature class it belongs to. Supports Matter, printed on a box, tells that buyer almost nothing.

Certified, and still not working where you put it

The same review lists specific failures rather than general grumbling. IKEA’s Bilresa remote and its Klippbok water detector do not work in the Google ecosystem. Alexa lacks leak-sensor support entirely, so a certified leak sensor has nowhere to report to.

Feature loss when a device is bridged is described as routine rather than exceptional. Button programmability is the standing example: whether single, double, triple and quadruple presses, or short and long presses, are available at all depends on which ecosystem is doing the controlling. The device supports them. Your platform may not expose them.

• A certification mark records what the device was tested against, not what your hub will do with it
• The failure is usually silent: the device pairs, then some capability simply is not there
• Bridged devices lose more than natively joined ones, and which capabilities go varies by platform

Over 750 products, counted generously

The headline number quoted for Matter device availability comes from a database listing over 750 products available or imminent. The same source is candid that the count includes country variants and colour or size differences, so the number of genuinely distinct device types is meaningfully smaller than the figure suggests.

There is a worse disclosure problem behind it. Vendors do not publish which version of Matter they implement. References to Matter are frequently buried in technical data rather than stated, which leaves a buyer with no way to determine what a device actually supports before opening the box. When neither side of the transaction will name a version number, the logo is doing marketing work rather than informational work.

Exposing a device is not the same as adding one

Philips Hue’s Bridge Pro is the cleanest example of a partial implementation sold as a whole one. Hue devices are exposed outward to Matter ecosystems, but third-party Matter devices still cannot be added into the Hue app. The bridge speaks Matter in one direction. A buyer who assumed a Matter bridge would let them consolidate their sensors inside Hue has bought the wrong thing.

The Aqara Hub M3 is the other case. PCWorld found the hub supports only select device types as a Matter controller and reported being unable to locate any canonical list of M3-compatible devices. It also does not work with third-party Zigbee devices directly and needs intermediate bridges for them. Aqara markets the M3 as an edge hub prioritising local control, yet when PCWorld unplugged the modem, attempts to control devices invariably timed out until the internet connection came back, and the supposedly offline-capable device-binding automations could not be activated at all.

The pricing context is worth keeping: PCWorld noted the M3 launched at $129.99 against the Hub M2’s then-price of $50.

Thread 1.4 became the only certification, and the meshes still split

Thread 1.4 was published in autumn 2024 and, since 1 January 2026, has been the sole border-router certification standard. Its three fixes address exactly the problems buyers complain about: standardised credential sharing so a 1.4 border router can grant temporary admin access through one-time passcodes, unified networks with proper IPv6 connectivity so devices from different vendors join one mesh instead of vendor islands, and better diagnostics for multi-router setups.

Adoption is where it stalls. Apple has shipped it, while Google, Amazon and Samsung were at varying stages of testing or planning. Among border-router implementations, SmartThings went first and IKEA followed. The user-visible symptoms have not gone away: devices disconnecting and reconnecting unpredictably, several Thread networks appearing in an app and refusing to consolidate, a newly added hub spawning yet another fragmented network, and extra battery drain as devices hunt for a parent router.

One measured upside is worth recording. Newly certified Thread sensors reach around two years of battery life, against about three years for comparable Zigbee sensors, with Aqara’s FP300 given as the example. Thread is not a battery disaster. It is a network that still needs supervision.

What we would buy on a Matter argument, and what we would not

The device we would specifically not buy as a Matter purchase is the Aqara Hub M3. It is a capable piece of hardware, but as the Matter brain of a house it fails the two tests that matter most: PCWorld could not establish which Matter devices it supports, because no canonical list exists, and the local operation it is marketed on did not survive the modem being pulled. Paying a premium for edge control that stops at the router is the wrong trade.

For everything else, buy on evidence rather than the logo. Ask the vendor which Matter version the device implements and treat silence as a warning, because silence is the industry norm rather than an oversight. If a bridge is involved, establish which direction it works in before assuming it will consolidate anything. If Thread is involved, prefer a border router certified to 1.4, since that has been the only certification available since January.

And accept the honest conclusion. Future-proofing is not something a badge can sell you in 2026. The specification is running ahead of every ecosystem that implements it, and the safest purchases are the ones that would still be acceptable if the Matter half never improved.

Sources: Connectivity Standards Alliance · matter-smarthome.de · Bitdefender · Hueblog · PCWorld