Does Frigate require a cloud subscription?

No, the core software is free and open-source under the MIT license with no per-camera fees. An optional paid Frigate+ subscription is available for custom models trained on your own labeled clips.

What hardware does Frigate need to run effectively?

You must provide your own hardware, including a decent CPU, an AI accelerator like a Coral TPU or supported GPU, and storage. Using a Coral TPU or GPU allows it to sustain over 100 detections per second.

How difficult is the setup process?

Setup is considered Docker and YAML heavy and is not beginner-friendly. Users should expect to spend time configuring Docker, go2rtc, camera substreams, and detector tuning before the system operates optimally.

Which home automation platforms does Frigate support?

Frigate offers native Home Assistant entities, MQTT events for NodeRed and OpenHAB, a built-in web UI, and an HTTP/WebSocket API for stream and event access.

Frigate

Open-source NVR that runs AI object detection locally on your security camera feeds.

Freemium· Free and open-source; Frigate+ custom-model subscription is paidVideoCustom object detection (Coral/YOLO/Frigate+ models)6.9 / 10

Visit website →

In short

Frigate is a self-hosted Network Video Recorder that ingests RTSP feeds and performs real-time AI object detection on local hardware to reduce false positives. It is best suited for privacy-minded homelabbers and Home Assistant users who want full control over their surveillance data without cloud subscriptions.

Best for

Pick Frigate if you want a privacy-respecting, AI-driven NVR that plugs into Home Assistant and runs entirely on your own hardware.

Skip if

Skip it if you want a plug-and-play cloud cam app or you don't have the patience for Docker, RTSP and detector tuning.

Frigate is a self-hosted Network Video Recorder built around real-time AI object detection. Instead of streaming footage to a cloud service, it ingests RTSP feeds from your existing cameras and runs inference on local hardware, classifying people, vehicles, animals and other objects with zone-based tracking. With a Coral TPU or supported GPU it can sustain 100+ detections per second, which cuts the false-positive flood you get from motion-only NVRs.

It's aimed at home automation tinkerers and privacy-minded homelabbers, especially the Home Assistant crowd, where it's the de-facto AI camera backend. The core software is free and open source under the MIT license; the optional Frigate+ subscription adds custom models trained on your own labelled clips for better accuracy on edge cases. There's no SaaS lock-in and no per-camera fee, but you bring the hardware: a decent CPU, an AI accelerator, and storage.

Integrations include native Home Assistant entities, MQTT events for NodeRed/OpenHAB, a built-in web UI, and an HTTP/WebSocket API for stream and event access. The trade-off is operational: you'll spend a weekend on Docker, go2rtc, camera substreams and detector tuning before it sings.

Editor's take

Frigate is the clear winner for self-hosted AI camera surveillance and the reason a lot of people buy a Coral USB stick in the first place. It rewards effort: budget a weekend for setup, but you'll never pay a Ring/Nest subscription again. Frigate+ is worth it once you hit the limits of the stock models.

— The AI Tool Bible editorial team

Pros

✅ Fully local processing, no cloud, no subscription required for core
✅ MIT-licensed open source with active community
✅ First-class Home Assistant and MQTT integration
✅ Real-time detection at 100+/sec with a Coral TPU

Cons

⚠️ Setup is Docker/YAML heavy and not beginner-friendly
⚠️ Best performance requires a Coral or GPU accelerator
⚠️ Frigate+ custom models are a paid add-on

Use cases

security-camera-nvrobject-detectionhome-automationprivacy-focused-surveillance

Frequently asked

Does Frigate require a cloud subscription?: No, the core software is free and open-source under the MIT license with no per-camera fees. An optional paid Frigate+ subscription is available for custom models trained on your own labeled clips.
What hardware does Frigate need to run effectively?: You must provide your own hardware, including a decent CPU, an AI accelerator like a Coral TPU or supported GPU, and storage. Using a Coral TPU or GPU allows it to sustain over 100 detections per second.
How difficult is the setup process?: Setup is considered Docker and YAML heavy and is not beginner-friendly. Users should expect to spend time configuring Docker, go2rtc, camera substreams, and detector tuning before the system operates optimally.
Which home automation platforms does Frigate support?: Frigate offers native Home Assistant entities, MQTT events for NodeRed and OpenHAB, a built-in web UI, and an HTTP/WebSocket API for stream and event access.