The r1o Guide

Download r1o.dmg, drag it to Applications, open it. That is the entire install. asmi, Hermes, the dashboard, model management — everything is bundled. No Homebrew, no pip install, no Docker.

On first launch, r1o scans your ~/Models directory and any HuggingFace cache automatically. If you already have MLX models downloaded, they show up instantly. Pick one from the Models tab, hit Deploy, and start chatting.

Tailscale creates an encrypted WireGuard mesh between your devices. Install it on your Mac and sign in — that is all. No port forwarding, no firewall rules, no cloud relay. Every device on your Tailscale network can reach your Mac directly.

Now open r1o on your iPhone or iPad. It discovers your Mac automatically over the Tailscale mesh. Chat from any device — the couch, the office, your commute. The model runs on your Mac; the phone is just a window into it.

Open r1o on a second Mac. If it is on the same Tailscale network, it joins automatically. The dashboard shows both nodes in the topology view — chip model, RAM, running models, everything visible from one screen.

Transfer models between nodes directly from the dashboard. Over Thunderbolt 5, transfers run at 5.5 GB/s. Over LAN, it falls back to rsync — slower, but it just works. Deploy a model to any node with one click.

Plug Thunderbolt 5 cables between your Mac Studios. macOS 26.2+ supports RDMA over these cables — direct memory access at Thunderbolt 5 link rates. This is how the cluster shares model weights at datacenter speeds.

The topology view in the dashboard lights up with neon cables showing active RDMA links. Model transfers jump from ~100 MB/s over LAN to 5,500 MB/s over RDMA. A 40 GB model transfer takes 7 seconds instead of 7 minutes.

Some models are too big for a single Mac. GLM-5.2 at 743 billion parameters needs over 400 GB of RAM just for the weights. No single machine can run it. JACCL splits the model across RDMA-connected nodes — you deploy, it figures out the sharding.

The Models tab in Mission Control is more than a file browser. Search by natural language — "fast coding model under 50GB" — and get ranked results with RAM fit estimation for every node in your cluster.

Download models directly from HuggingFace, delete old ones, transfer between nodes. The model registry tracks everything: size, quantization, architecture, community benchmarks, which nodes have a copy.

Hermes is the on-device AI agent that manages your cluster. It ships inside r1o — not a separate install, not a cloud service. Hermes speaks MCP and has access to every tool in the system: model management, cluster operations, your personal knowledge base.

Teach Hermes your workflows with custom skills. Skills are markdown files that describe procedures — Hermes reads them and executes the steps. Your conversations never leave your network.