Skip to content

Local models

Local models (/playbook) manages on-machine LLM runtimes: hardware scan, model download, and Ollama integration.

Naming

Local models (/playbook) is not the same as Playbooks (/playbooks), which are YAML automation workflows. See Playbooks.

Web UI (/playbook)

  • Scan CPU, RAM, GPU, and disk
  • List recommended models for your hardware
  • Pull and run models via Ollama (when installed)
  • Point Keprix providers at http://localhost:11434/v1

Configure as provider

  1. Install Ollama
  2. Pull a model: ollama pull llama3.2
  3. In Dashboard > Settings > LLM Providers, add Ollama or a custom provider with base URL http://localhost:11434/v1
  4. Select the model in chat

API

Action Endpoint
Hardware scan POST /api/playbook/scan
List models GET /api/playbook/models

Environment

OLLAMA_HOST=http://127.0.0.1:11434