Fregata Documentation

Fregata is a private, local-first network video recorder for macOS, built on the open-source Frigate NVR. Fregata runs object detection on the Apple Neural Engine via a custom Rust + CoreML detector, and decodes and encodes camera streams through VideoToolbox on the dedicated media engine — the CPU stays free for everything else.

Fregata ships as a signed, notarized .app bundle — drag to Applications, done. No Docker, no Linux VM, no terminal setup. macOS 13 (Ventura) or later, Apple Silicon (M1 or newer).

Install Fregata → Drag-to-Applications. macOS 13+, Apple Silicon.

Buy ($10) One-time. 12 months of updates. Yours forever.

GitHub ↗ The public mirror — issues, discussions, releases.

Looking for Frigate's docs?

These docs cover Fregata-specific topics — the macOS port, the CoreML detector, packaging, licensing. Everything else — every config.yml key beyond the detector, MQTT topic shapes, the HTTP API, zone geometry, semantic search, GenAI, face recognition, license plate recognition, PTZ tracking — lives in the upstream Frigate Docs ↗. The two projects share a config schema, so most Frigate documentation applies to Fregata unchanged. We don't mirror it; the upstream docs are excellent.

What you get

Apple Neural Engine

Object detection runs on the ANE, not your CPU or GPU. Lower latency, lower power draw, cooler fans. Bundled YOLOv9-tiny model out of the box; drop in your own ONNX model anytime, including Frigate+ models.

VideoToolbox decode

Hardware H.264 / HEVC decode and encode on the dedicated media engine. Multiple 4K streams without saturating cores.

Local-first

Recordings live on the Mac you choose. No mandatory cloud, no telemetry by default, no third-party access to your feeds.

One-time purchase

$10 unlocks a year of new versions. Already-released versions keep working forever, even after the license expires — only future releases require a renewal.

Start here

Installation System requirements, download, and the first launch.

Activation Enter your license key (or start a 30-day trial) and bind Fregata to this Mac.

Migrating from Frigate to Fregata Already running Frigate on Docker on a Mac? Step-by-step walkthrough for switching to a native install — config, recordings, integrations.

Add your first camera Point Fregata at an RTSP feed and watch the detection ratchet up on the menu-bar tray.

Once you're up and running

Detection tuning Masks, zones, per-object thresholds, ANE vs GPU, custom ONNX models, Frigate+.

Recordings & retention Continuous vs event-triggered, per-camera retention, disk-space budgeting.

Home Assistant Wire Fregata into the official Frigate HACS integration.

Performance Diagnose CPU spikes and detector latency; understand the ANE / GPU / CPU tiers.

Troubleshooting Common error messages and what to try, indexed by the exact string.

Fregata vs Frigate Exactly what the macOS port adds, removes, and changes — and where to read which docs.

Fregata vs Scrypted, Blue Iris, Synology Side-by-side comparison against the NVRs Mac users typically evaluate. Where each one wins.

Inside Fregata's CoreML detector Engine-room view: Rust dylib → ONNX Runtime → CoreML EP → ANE / GPU / CPU. How Fregata lands detection on the Apple Neural Engine.

Adaptive transcoding Fregata-exclusive: bandwidth-adaptive HLS for recordings and live view — works on weak LTE and 5G. Apple Silicon Media Engine, zero CPU.