Aqara G400 vs. G410: Why I’m now running TWO doorbells (and ditching Arlo)

TL;DR
I’ve finally moved away from Arlo and their constant subscription hikes, but my journey to a single-doorbell solution didn’t go as planned. I’m currently running both an Aqara G410 and the new G400 (PoE) side-by-side because neither one is a complete package.

My doorbell journey started with an Arlo doorbell. When I upgraded my outdoor camera from Eve Outdoor to the G5 Pro, I came across the Aqara G410 and decided to switch—mainly to move away from Arlo’s ever-increasing subscription costs.

After installing the G410 (luckily in a different position, since it’s wider and couldn’t fit in the original Arlo spot), I removed the Arlo doorbell. Overall, the G410 works great—HKSV, SD card, NAS support, all solid. I’m also fine with the 4:3 format.

However, one major drawback: there’s no “call to phone” function when someone presses the doorbell, which is a key feature for us. So I ended up reinstalling the Arlo just for that function. For a period, I was running both the G410 and Arlo side by side.

Later, I added a T2 relay to control my electric drop bolt. With that setup, I was able to use the G410’s local face recognition to automatically unlock the door. (Yes, it’s 2D recognition, but since this is for an outer door, security wasn’t a big concern—this was purely for convenience.)

When the G400 was announced with “call to phone” support, I thought I could finally consolidate everything into a single device and replace both the G410 and Arlo.

Installation went smoothly, but I quickly realized:
:point_right: The G400 does not support local face recognition.

To use face recognition, I had to subscribe to Home Guardian. Not ideal since I’m trying to move away from subscriptions, but I was willing to accept it—as long as pricing remains reasonable.

Unfortunately, the issues didn’t stop there.

  • In low light, the G400 image is darker than the G410, which affects face recognition accuracy
  • This can be partially mitigated by adjusting brightness or using the G5 spotlight
  • But the bigger issue: cloud-based recognition is significantly slower

I experienced delays of up to 5 seconds (even with PoE). There were many instances where I had already entered and closed the door before recognition triggered—only for the door to unlock again afterward.

My wife and I had already gotten used to the instant response of local face recognition, so this delay became a deal breaker. I don’t mind paying for a service, but not at the cost of usability.

My Current “Frankenstein” Setup

  • Cancelled Home Guardian subscription
  • Reinstalled the G410 alongside the G400
  • G410: used purely for local face recognition and auto-unlock
  • G400: used for everything else (including “call to phone”)

Conclusion:
This isn’t a clear “better vs worse” situation—both devices solve different problems.

  • G410 → better for fast, local automation
  • G400 → better for feature completeness (e.g., call to phone)

The main consideration is whether your setup prioritizes real-time response or integrated features via the cloud. In my case, I ended up needing both.

3 Likes

Kind of my issue as well

Very disappointed with the G400. Two issues that I’ll keep short.

1.) When paired with HomeKit and aqara, the video quality is so reduced it looks like 480p or worse, very grainy.

2.) Detection is terrible. My 6 year old circle view doorbell I needed to set activity zones because it would notify me every time someone walked on the sidewalk or drove by on the road. This camera I have no such zone so it should see everything, yet it catches nothing. I does catch the people coming up to my door, but not much further than that.

I was excited for this doorbell and then disappointed within the same day I got it. I think it’s beyond time for me to just throw in the towel on the subpar half a decade ive spent with HomeKit enabled devices.