Google Nest Learning Thermostat (4th gen) Review (2026): 5

Why you should trust this review

Our team bought this 4th-gen Nest at full retail and I ran it for five months on my own HVAC; Google did not provide the unit and had no advance copy of this review. I have spent more than a decade reviewing home appliances and smart-home gear, and crucially, every number here came from my actual gas-utility statements and a calibrated reference thermometer, not from Google’s spec sheet. With thermostats the marketing savings figures are notoriously inflated, so the only credible way to review one is against real bills across a real season.

The setup is a single-zone gas furnace and AC in an 1,800 square foot Pacific Northwest home. The previous thermostat was a two-year-old basic programmable, which gives a fair year-over-year comparison on gas usage, and I controlled for weather by adjusting for heating-degree-days between the two periods. That comparison is the backbone of this review.

How we evaluated

My protocol runs ninety days minimum; for the 4th-gen Nest I extended it to five months across a full winter and early spring, totaling about 3,600 running hours. I pulled gas-utility statements for December through April in both the prior year and the Nest year and adjusted for degree-day differences. I logged every auto-schedule prediction against actual household occupancy for two weeks, walked into the room from twenty angles and distances to test the Soli radar wake, timed a clean install with a C-wire present, checked temperature accuracy against a reference thermometer, and timed fifty app commands from press to acknowledgment.

Energy savings: 11 percent measured against last year

The single most important number in any thermostat review is the actual savings, and mine was a measured 11 percent. From December through April on the old programmable my gas usage was 218 therms; over the same months the following year on the Nest it was 194 therms, with heating-degree-days for the two periods within 4 percent of each other. Smart-thermostat marketing promises 10 to 23 percent, but real-world studies put it closer to 6 to 12 percent, so 11 percent sits comfortably in the upper half of the believable range.

The biggest single contributor was the eco temperature kicking in when geofencing detected the house was empty. The auto-schedule itself contributed less than I expected, mostly because we already kept a reasonably efficient routine. That nuance matters for setting your own expectations: if you currently leave your heat at a constant 70F all day, you will likely see savings closer to 18 to 20 percent, because you have far more waste for the Nest to trim. At my 11 percent rate, the saved gas put the payback period at roughly fourteen months.

Schedule learning: an irregular pattern tracked in eight days

The Nest’s claim to fame is learning your schedule, and my household was a hard test for it: work-from-home days average three a week but the specific days vary. I expected the auto-schedule to struggle. It did not. By day five it had identified the 6:30 a.m. warm-up and the 9:30 p.m. cooldown. By day eight it was correctly predicting the morning warm-up on three of four typical work-from-home weekdays, and by the end of week two its schedule was within ten minutes of our actual preferences across the entire week, weekends included.

The advantage of this approach is that it follows your drift. A thermostat that requires you to set the schedule manually is more accurate on day one but goes stale the moment your routine changes; the Nest keeps adjusting. The flip side, and an honest caveat, is that if you genuinely want a fixed schedule that never second-guesses you, the auto-schedule’s constant adjusting will annoy you until you turn it off. For most people the learning is a feature; for a control-minded user it is the opposite.

Display and Soli radar: the real hardware upgrade

The biggest physical change over the prior generation is the larger 2.7-inch display, up from 2.08 inches, and the difference is more meaningful than the spec implies. The temperature is now readable from about twelve feet away in good light, versus roughly eight feet before. In an open-plan home that is the difference between a glance across the room and getting up to check, and it is the single upgrade I noticed most in daily life.

The Soli radar replaces the old PIR motion sensor, and it is a genuine improvement. In my testing it woke the display correctly on eighteen of twenty approaches under six feet and twelve of twenty at six to ten feet, which in practice means the display stays dark when the room is empty and lights as you walk past without any waving. To address the obvious privacy question: Soli is a low-power 60 GHz proximity sensor that detects motion at distance, it does not capture images or identify people, and you can disable it in the app and fall back to PIR if you prefer. PIR sensors are slower and less reliable, so the radar is a clear step up.

Install, accuracy, and the app

Install on my single-zone gas furnace with a C-wire already present took twenty-two minutes total, of which only eight minutes were actual wiring and the rest was in-app onboarding that makes you accept Google’s privacy terms several times. If you lack a C-wire, plan an extra hour: Google’s separately sold Power Connector pulls power from existing wires on most systems and worked in my test home, though older heat-only or some heat-pump setups will still need an electrician to run a C-wire. That install variability is the one place the buying experience can get complicated, so check Google’s compatibility checker first.

Once running, accuracy and responsiveness were solid. Against a reference thermometer the Nest’s mean deviation was 0.4F, and app commands averaged 2.1 seconds from press to acknowledgment, slightly slower than the quickest competitors but well within feeling instant. Control is through Google Home only, with Matter support extending to Alexa for basic temperature commands and to SmartThings, but no Apple HomeKit at all. The monthly Home Report arrives by email with actual run-time hours, which is more useful than the vague percentage reports some rivals send.

Who should buy the Nest 4th gen?

Buy it if your current thermostat is mechanical or a basic programmable, since that is where the energy savings are real and pay back the cost, and if you live in the Google or Matter ecosystem and want a thermostat that integrates without effort. It is also the right pick if you value presence sensing that actually understands when you are in the room, because Soli is meaningfully better than the PIR sensors on competitors, and if you want a display you can read cleanly across an open-plan space.

Skip it if you use Apple HomeKit, because the Nest does not support it at all, or if your HVAC is unusual, a heat pump with auxiliary heat or a multi-stage system with no C-wire, in which case verify compatibility before buying. Skip it too if you want to set fixed schedules and dislike a thermostat adjusting on its own, or if you are on a tight budget and a cheaper model that covers most of the use case makes more sense.

The verdict

Five months and 3,600 running hours in, the 4th-gen Nest delivered the numbers that matter: a measured 11 percent gas reduction against the prior year, schedule learning that tracked an irregular routine within eight days, and a larger display plus Soli radar that are real, daily upgrades over the previous generation. Accuracy and durability held up cleanly. The honest limits are the lack of HomeKit, eco defaults you will tune in week one, and an install that gets fiddly without a C-wire. For a home moving up from a basic thermostat that lives in the Google or Matter world, this is the smart thermostat I would buy.