Frigate
Camera recording with object detection, usually paired with Home Assistant.
The heaviest app on this list. Camera count, detector hardware and recording retention decide the machine, and recordings are user data.
Planning figures used by the calculators: RAM 4 GB (our estimate), CPU 2 cores (our estimate), storage 20 GB (our estimate). These cover the application itself. Your own photos, films, documents and recordings are entered separately as data storage.
The table lists every figure we hold, what kind of claim it is, and the fragment of the archived document it came from.
Source: Frigate recommended hardware, retrieved 2026-09-08, archived as document 8 on the sources page.
Add Frigate to a stack or read how we label a number.
| property | value | basis | evidence | document |
|---|---|---|---|---|
| cpu plan cores | 2 cores | our estimate | Our reasoning: No published figure. Assumes a supported detector or iGPU does detection. Detection on the CPU alone needs considerably more. | No document, our reasoning only |
| cpu requirement | Intel CPU with AVX and AVX2, able to run Debian | vendor documented | Quoted from the archived document: Anything with an Intel CPU (with AVX + AVX2 instructions) and capable of running Debian should work fine. | Document 8, retrieved 2026-09-08 |
| detector | A detector is expected, not optional | vendor recommended | Quoted from the archived document: Frigate is designed around the expectation that a detector is used to achieve very low inference speeds. | Document 8, retrieved 2026-09-08 |
| ram plan gb | 4 GB | our estimate | Our reasoning: No published figure. Decoding several streams plus detection and the recording cache; we plan 4 GB for a few cameras. | No document, our reasoning only |
| storage plan gb | 20 GB | our estimate | Our reasoning: Application and cache only. Recordings are entered as data storage and dominate the total. | No document, our reasoning only |