Model Deck local gateway

The need

Switching between local AI and several provider interfaces fragments prompt work and credentials.

Why it was developed

Provide one local interface and gateway with Windows-account-protected provider keys and a local Ollama option.

How to use it

1. Install Python 3, a browser and Ollama. Download a suitable model once using Ollama, then start its local service.
2. Run Launch Model Deck.cmd; open http://127.0.0.1:4173 if the launcher does not open the browser.
3. Choose a local Ollama model for the route that has no mandatory cloud inference charge.
4. Add cloud-provider credentials only when desired; usage budgets displayed by this app are planning estimates, not authoritative provider billing balances.

How it can save time or money

Reduces application switching and repeated provider setup. Local Ollama routing avoids mandatory per-request cloud fees for supported local tasks.

These are potential benefits, not measured savings. Include setup, maintenance, review and local operating costs when comparing your own workflow.

Current scope & limits

Cloud providers may charge independently. Cached models and encrypted provider-key data are excluded. Some launch helpers reference the original D: installation.