Install on Linux
This installs the local gateway on Linux amd64 or arm64. Confirm you meet Requirements first.
1. Install and start Ollama
Install Ollama, start it, then pull the model:
ollama pull gemma3:4b
Use Ollama 0.33.3 or newer. Older versions are significantly slower and unsupported.
2. Find your machine's LAN address
The camera needs a specific private IPv4 address on the same network:
hostname -I
If several addresses are listed, pick the one on the same subnet as the camera. If you use a VPN or multiple network interfaces, choose carefully; setup does not guess this value. Hostnames and IPv6 are not supported.
3. Download and run the installer
Download install-anava-local.sh from the versioned release directory shown in your Anava portal. The portal shows the exact verified version string; use it in place of VERSION below.
sh ./install-anava-local.sh --version VERSION --advertise-host YOUR_LINUX_LAN_IP
The installer:
- Downloads the matching Linux amd64 or arm64 archive from
https://storage.googleapis.com/anava-acaps/local-gateway/vVERSION/. - Verifies the download against
SHA256SUMS. - Installs the executable to
$HOME/.local/bin. - Initializes
$HOME/.config/anava-local. - Installs and starts the gateway as a user service (systemd).
The installer never requires root access and never modifies your existing Ollama installation or models.
4. Confirm the service is running
$HOME/.local/bin/anava-local-gateway service status --state-dir "$HOME/.config/anava-local"
5. Locate the connection file
The installer creates ~/.config/anava-local/camera-connection.json. This file is a bearer credential: it contains the HTTPS endpoint, the fixed anava-local model alias, a shared secret, and a private CA certificate. Treat it like a password. Do not email it or commit it to a repository.
Dedicated hosts
If you run the gateway on a shared or GPU-loaded Linux host, expect slower and less predictable latency. Use a dedicated machine when you need consistent response times.
Next step
Related: anava.ai/edge/local