> ## Documentation Index > Fetch the complete documentation index at: https://docs.bricks.tools/llms.txt > Use this file to discover all available pages before exploring further. # Use Buttress for inference > Offload LLM and speech-to-text generators to a Buttress server on your LAN BRICKS Foundation can offload LLM, MLX, and speech-to-text inference to a [BRICKS Buttress](/buttress) server on the same LAN. The brick's generator API stays the same; the device transparently delegates the work to the server when it would run faster (or fit at all) there. ## Requirements * BRICKS Foundation 2.25 or later * A Buttress server reachable on the LAN — see [Buttress installation](/buttress/installation) * The server is bound to the same workspace as the device — see [Workspace binding](/buttress/workspace-binding) ## How it works 1. When the device's workspace has at least one bound Buttress server, the launcher starts a discovery manager and obtains a workspace-scoped JWT. 2. The manager listens for UDP announcements on port `8089` and pools every server whose `serverId` is on the workspace's bound list. 3. Each LLM, MLX, or STT brick reads its **Buttress (Remote Inference)** group and either picks a server automatically or uses a manually configured URL. 4. The capability comparison decides whether to run locally, remotely, or either; the strategy you pick on the brick decides how that recommendation is interpreted. If the launcher can't reach a server (no LAN, server offline, workspace mismatch), it falls back to local execution — unless you explicitly turned that off. ## Configure offloading on a brick In **BRICKS Controller** > **Config Editor**, open any LLM or STT brick. The **Buttress (Remote Inference)** property group appears under **Connection**. | Field | Default | Effect | | ----------------- | ----------------- | ---------------------------------------------------------------------------- | | **Enabled** | `false` | Turn Buttress on for this brick | | **Auto-discover** | `Auto` | `Auto` finds a server via UDP; `Manual` uses the URL field | | **URL** | empty | WebSocket URL when Auto-discover is `Manual` (e.g. `ws://buttress.lan:2080`) | | **Strategy** | `prefer-buttress` | How the device picks between local and remote | | **Fallback** | `no-op` | What to do when Buttress is enabled but unavailable | ### Strategies | Strategy | Behavior | | ----------------- | -------------------------------------------------------------------------------------- | | `prefer-buttress` | Always send work to Buttress when a server is available. Local hardware is not probed. | | `prefer-local` | Run locally if the device has enough memory; otherwise fall back to Buttress. | | `prefer-best` | Compare scores and run on whichever side is faster. | `prefer-buttress` is the default because Foundation devices that opt into Buttress almost always do so because the local hardware is not the fastest path. ### Fallback | Fallback | Behavior when Buttress is enabled but unreachable | | ----------------- | --------------------------------------------------------------------------- | | `no-op` (default) | The brick does nothing locally — no model is downloaded, no completion runs | | `use-local` | The brick falls back to local execution exactly as if Buttress was off | Pick `use-local` if you want the brick to keep working when the LAN drops; pick `no-op` if you would rather see a clear failure than silently consume battery on a model the device can't handle. ## When the workspace changes If the device's workspace changes — for example, an admin reassigns it from BRICKS Controller — the launcher: 1. Stops the active Buttress manager and closes any open WebSocket connections. 2. Discards the cached access token. 3. Starts a new manager with the new workspace's bound-server list and a freshly issued token. In-flight generators that were authed against the old workspace error out cleanly rather than entering an infinite reconnect loop. ## Audio uploads (STT) Speech-to-text transcription needs the audio file on the server. The brick uploads to `POST /buttress/upload` over HTTPS, and the server stores the file in the temp directory configured by `[server] temp_file_dir` (default `/.buttress`). After transcription, the file is auto-cleaned along with the rest of the session's temp files. ## Troubleshooting | Symptom | Likely cause | Fix | | --------------------------------------------- | ------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------- | | Brick logs `no LAN provider is registered` | Device is not associated with a workspace, or no servers are bound to that workspace | Bind the device, or run `bricks buttress bind` to pair a server | | Brick logs `no '' endpoint yet` | UDP discovery hasn't returned a server within 10 seconds | Check that the server is on the same subnet and the `[autodiscover]` block isn't disabled | | WebSocket closes with code `1008` | The device's token doesn't match the server's bound workspace | Confirm both sides are on the same workspace; restart the launcher | | Brick "stops working" after enabling Buttress | Fallback is set to `no-op` and no server is reachable | Switch fallback to `use-local`, or fix LAN connectivity | | Unknown generator id error | Server restarted or evicted the loaded model | The brick recovers automatically by re-initializing the generator on the next call | ## Related What Buttress is, when to use it, and how the system fits together. How servers and devices end up on the same workspace.