A day after the initial release , both Hailo roles get version 0.2.0 - with two notable jumps: selectable package sources (Raspberry Pi repository or Hailo vendor) and a full web-UI delivery built around Open WebUI and an optional Caddy TLS frontend.
head1328.hailo- driver and runtime for the Hailo NPUhead1328.hailo_ollama- Hailo Ollama LLM server with Open WebUI and Caddy reverse proxy
head1328.hailo#
The base role installs the Hailo driver and runtime. Supported hardware:
| Device | Variant | TOPS |
|---|---|---|
| AI HAT+ (Hailo-8L) | h8l | 13 |
| AI HAT+ (Hailo-8) | h8 | 26 |
| AI HAT+ 2 (Hailo-10H) | h10 | 40 |
New in 0.2.0:
- Selectable source:
hailo_source: raspberry(Raspberry Pi OS apt repo) orhailo_source: vendor(packages fromdev-public.hailo.ai) hailo_vendor_versionandhailo_vendor_base_urlvariables for pinning vendor packages- Automatic cleanup of conflicting packages when switching between
raspberryandvendor - More robust apt state handling across source switches
- Dedicated molecule scenario for the vendor variant, including idempotence tests
head1328.hailo_ollama#
Builds on top of the driver role and adds an LLM server for the Hailo-10H NPU. Containers run rootless via Podman Quadlets.
New in 0.2.0:
- Selectable source: Hailo GenAI Model Zoo from Raspberry Pi OS or directly from the Hailo vendor
- Open WebUI as the chat frontend (enabled by default, pinned to
0.8) - Optional Caddy reverse proxy with TLS
- iptables PREROUTING redirect for privileged ports, persisted across reboots via
iptables-persistent - Caddy workarounds for Open WebUI 0.8 (
/configstub,/api/*/0path rewrite) - Open WebUI title generation and follow-up suggestions disabled by default (NPU limitation: one request at a time)
OLLAMA_HOSTenv var in the systemd unit for hailo-ollama 5.3.0 compatibility- Instructions for importing the Caddy root CA into the browser
What It’s For#
Anyone who wants to run a Raspberry Pi 5 with AI HAT+ 2 as a small local LLM server - no GPU, no cloud account required. Combined with head1328.ivly
you get a fully self-hosted back-office environment with LLM support.
Installation#
ansible-galaxy role install head1328.hailo
ansible-galaxy role install head1328.hailo_ollamaLicense#
Both roles: AGPL-3.0-or-later.
Acknowledgments#
- Hailo - NPU hardware and SDK
- Ollama - LLM server
- Open WebUI - chat frontend for Ollama
- Caddy - reverse proxy with automatic TLS
- Podman - daemonless container runtime
- Ansible Galaxy - role registry
Found this helpful?
Consider supporting via:
