I finally got around to updating ESPHome to 2026.1.x. Another nearly flawless update, one minor breaking change I easily adjusted for.
But the really remarkable thing was that, for the fourth monthly update in a row, each of my devices is using slightly less memory than before. And most of those are old 8266 chips which many here would consider obsolete.
I want to recognize the ESPHome developers for keeping code efficiency as a priority, even on old platforms.
I know that in the development world, “keeping the lights on” isn’t very sexy. All the glory goes to those who develop flashy new things for bleeding-edge technology. It’s easy to disregard old hardware and not worry about keeping things lean.
My thanks and appreciation go out to the whole ESPHome team. You are setting an example that many other developers should aspire to.
I agree here. While I’m not buying esp8266 devices anymore, we have lot of them still around. Also, newbies watch tutorials and go with the hardware used there.
If we look at Arduino users, most of them start with ancient 8bit classic Uno/Nano, not with Uno Q.
To me it’s no sense to buy esp8266 at the moment but that’s how it goes.
I agree. Definitely fulfills one of Open Home core values, of sustainability. Mega-corporations want us to throw out the old stuff, even though it’s still perfectly serviceable, but ESPHome team keeps it usable. Kudos!
I have almost 100 ESPHome devices here. 90% of them on the Wemos D1 Mini (ESP8266). (Only one is ESP32). The Wemos D1 Mini satisfies every requirement of the gadgets I make, and they are significantly cheaper than any ESP32.
Not only usable but they even claim a faster, leaner and more durable experience in 2026 with our 10+ year old MCUs
In 2026, ESPHome is faster, leaner, and more durable. It respects the hardware it runs on and the users who already invested in it.
Taking sustainability seriously
Privacy, choice, and sustainability are not abstract ideals. They show up in practical ways, in how long devices remain usable, how often hardware needs to be replaced, and whether users are forced to discard perfectly good devices just to keep their software up to date. We chose sustainability through efficiency.
Our home hosts dozens of esp82xx based devices, most of them have between 4 and 8 years of usage already and all of them going strong thanks to ESPHome
I used to build mostly everything around d1 minis but nowadays the natural successor surely is a esp32-c3 supermini which has more usable pins and an even smaller footprint (less than half as an d1 mini) while maintaining the same low price as d1 minis, typically around $2 per unit when sourced directly from the country of origin
Nice! I somehow missed that. This goes a long way toward restoring my faith in the Open Home Foundation.
I started with HA mostly because I had been looking for something to try out on those cool new Raspberry Pi things I’d been reading about. I bought the flagship of the fleet: RPi 3B+. It ran HA with room to spare.
This year I finally retired it. Not because it wasn’t still working as well as the day I bought it. I scrapped it because HA had moved on. I had become self-conscious about even mentioning it here. Afraid I’d be scoffed at. The consensus seemed to be “Throw that thing away! Go out and buy new hardware every couple of years!”
OK, maybe that’s an exaggeration. But it’s how I felt. HA developers assume everyone is running robust hardware, and will simply replace it as the code base grows. Maybe that’s unavoidable. Which is why it’s so refreshing to see respect for old hardware, sustainability and code efficiency from the ESPHome team.
I was using Home Assistant when everything was in one single configuration.yaml file. Everything was yaml- there was no GUI. If I wanted a sensor or control I had to make it using either an ESP8266-01 or a Wemos D1 Mini.
I also have several devices in hiatus (xmas and Halloween, for example) and not currently online.
I bought ten of these because it seems inevitable that I will be dragged kicking and screaming into ESP32. (For example, WLED will no longer support any ESP8266 board in a few months).
But my experiments with the supermini have been, character building. First tests using ESPHome, most would not connect to my WiFi AP even though it was only 1M from the AP. Arduino code compiled fine, but it too would not connect with the AP. ESPNow worked most of the time because it operates at the MAC layer (layer 2). But it was unreliable. I was able to install WLED on one supermini, but like the others connecting to WiFi was slow.
That’s a fair point I reckon and have observed that trajectory.
I also upgraded a few times through pi’s until I landed on HA running on an old laptop (It was quite satisfying to reuse it as it was collecting dust). But to be fair in my instance, I was asking HA to do more and more, including running some quite resource intensive add-ons (or should I say apps now).
I keep a pi for the odd occasion I need one for a task and I think another got “handed down” to a friend / new HA entrant.
You’re looking at this wrong. If all you ever expect from Home Assistant is to control a few lights and monitor a few sensors, then the Raspberry Pi4 is quite sufficient and will be for a long time. But you have gone further down the rabbit hole, and your needs have outgrown the Raspberry.
Fair enough. Tech moves on. But I really had kept a very lean HA environment on my RPi. I would have been content with that 3B+ for much longer, had not each version of HA crept a little farther up on the memory utilization chart. I never got much beyond single-digit CPU utilization on that thing when I had to scrap it. I suspect the same downward spiral will happen with the RPi 4.
These days, the RPi line is priced out of the market IMHO. No offense intended to anyone who still likes them. Just that I think I can do much better for the same money.
I’m excited to start using some of that extra CPU, memory and storage I have on the old laptop I’ve migrated my HA environment to. Now I can play with some of the add-ons (sorry, apps) out there. I’ve already tried some I never even knew existed.
Back on topic, I’m glad I don’t have to go through this same sort of thing with my ESP8266’s. I remain very impressed with the devs for keeping ESPHome lean and efficient.
Might be good idea, today I got my first dead Esp over 9 years, silent suicide in first days of use with no symptoms or warning signs… (C3 supermini)
I have many ESP8266 devices at home, such as the light switch module, which is not easy to replace. I would like to ask if the ESPNow function in ESPHome will support the ESP8266 chip in the future. I am looking forward to it. These switch modules are running well and there is no reason to replace them.
