PaletteAI Inference Launchpad How-to Guides
How-to guides get a specific job done on a running appliance. They assume you know what you want to accomplish and give you the steps to do it without teaching background concepts.
Contents
| Guide | What you do |
|---|---|
| Install the Appliance | Flash the installer ISO, boot the hardware, and bring up the appliance console. |
| Manage Cluster Infrastructure | Index the Local UI tasks for day-two cluster infrastructure operations. |
| Upgrade the Platform | Upload a newer content bundle from Artifact Studio and apply Update in Local UI. |
| Deploy a Model | Deploy an LLM, choose which nodes run it, and verify it is serving. |
| Replace a Model | Remove a model from a node, then deploy a newer version or a different model. |
| Upload a Model | Download a model on a jumpbox and upload it to the appliance. |
| Bring Your Own Model | Author metadata for a model that is not certified, then upload and deploy it. |
| Configure Semantic Routing | Set the Complexity threshold, author category rules, override both per client, and turn on Decision recording. |
| Set the Thinking Directive for a Tier | Choose off, on, or an effort level per tier on the Tier Map. |
| Enable Vision Preprocessing | Deploy a vision model and turn on image-to-text preprocessing for a text-only model. |
| Create a Client | Create a client and issue its first API token. |
| Generate an API Token | Create an API token that clients use to authenticate to the appliance. |
| Set and Manage Client Quotas | Set, edit, raise, and remove a client's request, token, and cost limits. |
| Manage a Client's Model Access | Route a client to models and allow it to reach external models. |
| Register an External Inference Endpoint | Register an OpenAI-compatible host, authorize a client, and route to its models. |
| View Token Usage | Find token usage by model and by client, and open the metrics dashboards. |
| View Client Usage | View quota utilization, historical consumption, and per-token usage. |
| Revoke or Delete a Client | Find expired keys, revoke a token, or delete a client. |
| Use Claude Code | Connect Claude Code to the appliance so a local model serves each request. |
| Use Cursor | Connect Cursor's Ask mode to the appliance through a GPT model the appliance serves. |
| Use OpenAI Codex | Connect the OpenAI Codex CLI to the appliance with a custom model provider. |
| Use OpenCode | Connect the OpenCode terminal agent to the appliance through a custom provider. |